Don't rewrite your application, extend it (12/18/1990)
Joel West, "Don't rewrite your application, extend it," MacWEEK,
v4, n42 (Dec 18, 1990):38.
If you haven't found an application that meets your organization's needs,
pause a minute before diving into the quagmire of custom application development.
In most cases, it's better to extend existing commercial applications using
external code resources.
Such externals (sometimes referred to as plug-ins) allow you to build a
custom environment for your users without having to write a sophisticated
application from scratch. The ability to write new C or Pascal code -- to
share data with workstations and mainframes, for example -- provides extensibility
and integration capabilities far beyond the speed and flexibility of macros.
We usually think of databases when we consider external code resources,
but externals also are becoming a standard feature of graphics programs,
word processors and spreadsheets.
Range of options. The capabilities and implementation of external
code resources vary by application type and by the proprietary external
formats of application vendors. The most flexible formats let you add almost
any function to an application, such as manipulating text strings, performing
complex calculations, displaying a custom user interface or playing sounds.
The best known of these externals are the external commands (XCMDs) and
external functions (XFCNs) of HyperCard. These provide a mature, coherently
designed interface (not something hacked together by a programmer one weekend)
that are better documented than all other externals put together. If you
don't like the Apple (now Claris) manuals on XCMDs, you can find plenty
of books or magazine articles providing tips and examples. Or you can hire
the services of a local HyperCard consultant.
The disadvantage, of course, is that XCMDs are an extension of HyperCard,
which can be made to do just about anything but is rarely the most appropriate
foundation for a business application.
Still, similar external code resources can be added to many database and
spreadsheet applications, providing a wider range of options for information
retrieval and analysis.
The filter approach. Other applications offer extensibility via a
single well-defined function, typically import or export filters for other
data formats.
Such import and export filters are used by graphics and word processing
programs to convert files produced by other applications to their native
format. They also are often used for importing images from a peripheral
directly into the application, such as reading a gray-scale image from a
scanner into a photo-retouching program.
The best-known such format is Claris XTND, supported by Claris' CAD, MacWrite
II and FileMaker Pro. XTND also is being licensed and used by other application
vendors, notably Symantec Corp.
Although other approaches to data conversion exist, external filters provide
a seamless integration of unusual data formats that is as easy to use as
a standard format.
Programmer beware. For some common functions (such as converting
data to IBM PC formats), the externals you need may already be available
as commercial, off-the-shelf software packages.
But if you plan to write your own externals, be wary of the quality of the
documentation, which varies from vendor to vendor. With the exception of
XCMDs, little documentation is available for writing externals except from
the application's publisher. Check the documentation and write a few simple
externals before committing significant resources to a particular application.
In the future, externals will not only be more common but will work with
multiple applications from different vendors. By sheer market presence,
the Claris XTND format has the best prospect of becoming such a multivendor
standard.
Your system may not support virtual memory (11/20/1990)
Joel West, "Your system may not support virtual memory," MacWEEK,
v4, n40 (Nov 20, 1990):36.
When System 7.0 makes its official appearance, users will find its virtual-memory
feature a main attraction. But its usefulness will depend initially on hardware
and software.
As popularized on DEC minicomputers using VAX/VMS and Berkeley Unix, virtual
memory effectively increases the amount of RAM by tapping into free hard
disk space. This means that you will be able to run more applications (or
allocate more meory to each) without having to buy more SIMMs (single in-line
memory modules).
Instead, a Mac equipped with 2 Mbytes can simulate an 8-Mbyte machine, if
you're willing to dedicate 8 Mbytes of disk space to hold the virtual storage.
If an application tries to acces a memory location that's not in RAM, the
system will automatically read a group of adjacent memory locations (a "page")
from disk.
The hardware catch. This scenario assumes that you have purchased the right
Mac, however, since virtual memory will be available only on those Macs
with a processor or coprocessor that supports demand-paged virtual memory.
The Motorola 68030 used in the SE/30, IIx, IIcx, IIsi, IIci and IIfx fully
supports the machine-language instructions needed by Apple to implement
virtual memory; the 68000 (found in the Mac Plus, SE, Classic and Portable)
does not. The 68020 used in the Mac II and LC does not by itself support
the instructions but will with the addition of a 68851 PMMU (paged memory
management unit). the Mac II comes with a socket for a 68851, as do many
accelerators for the SE and Plus; the chip itself costs less than $200.
The LC is more problematic. It doesn't come with a PMMU socket, and Apple
hasn't announced plans for an add-on that would include one. Third parties
have announced cards that include a 68851, but using one of those in the
LC's single slot will preclude using Apple's IIe emulation card.
The software factor. Even if you have the right hardware, the benefits of
virtual memory also can be limited by software incompatibilities.
Most applications have been written to run the 64-bit (16 Mbytes) addressing
space designed in the original Mac's 68000. Since some of that 16 Mbytes
must be reserved for ROM, I/O and expansion slots, an application running
in 24-bit addressing mode is guaranteed only 8 Mbytes of addressable memory
(on most CPUs).
Applications that can run in 32-bit mode, however, will be able to access
up to 1 Gbyte of virtual memory if you reserve enough disk space to hold
the corresponding disk image. Unfortunately, to run in 32-bit mode, the
Mac's ROM also must be 32-bit clean, which is not true of the pre-1990 Macs.
Apple may patch or upgrade the ROM of the SE030 and older modular Macs to
make them 32-bit clean, but for now only the IIci, IIfx and IIsi can be
run in 32-bit mode.
Performance. Even applications that run under virtual memory will display
varying degrees of improved performance. Many word processors, for example,
probably won't benefit greatly from virtual memory, since they have been
designed to read their documents from disk a little at a time.
Many other applications will benefit immediately, allowing you to simultaneously
run a CAD program and image-processing software, for example, on a 2-Mbyte
machine.
But some applications not designed for virtual memory have a simple- minded
memory-management scheme that assumes all information is available in memory
at all times. these programs use memory in a way that causes the virtual-memory
software to spend most of its time reading pages from disk. Such a program
will be inefficient, particularly if it (or another application) is trying
to do computations in the background.
Even with these problems, the System 7.0 virtual-memory software is a cheap
enhancement for high-end Macs that will become even more useful as Apple
and third-party vendors iron out the remaining kinks.
New low-cost Macs: how will they fare? (11/6/1990)
Joel West, "New low-cost Macs: how will they fare?" MacWEEK,
v4, n38 (Nov 6, 1990):40.
The new trio of low-cost Macs marks a major shift for Apple. But what do
the machines presage for the Macintosh a year from now?
The message from Apple is that the future is sound, color and expandability.
The machine that epitomizes this future is the LC. It's the only machine
likely to have an impact across all markets, since the Classic is pretty
much limited to education and home use, while the IIsi primarily appeals
to business.
Machine of the future. The LC, which has the most important features of
the Mac II of three and a half years ago at a third of the price, also offers
sound input. Now that user-generated sound is available, application developers
will begin experimenting with both sound input and output, such as audio
annotation to replace the current Post-it note metaphor.
With sound input; sound output; expandability to 16-bit color via Apple's
video expansion SIMM (single in-line memory module); and the 32- bit, 68020
processor, the LC is also Apple's lowest-cost multimedia platform.
Although Apple seems fascinated with multimedia -- particularly animation
and sound -- this new technology will have a limited role in the general
user community, at least in the short run. Multimedia cannot supplant desktop
publishing in significance, so long as a presentation requires schlepping
a $2,500, 50-pound computer system.
Macs in education. The one major area where multimedia has proven advantages
is in training, as evidenced by the strong demand for courseware applications.
This meshes quite nicely with the LC's most strategically important role,
converting Apple's K-12 base from the obsolete Apple II to its mainstream
Macintosh product line. In addition to the LC's courseware potential, once
Apple ships the IIe emulator card, aggressive pricing of floppy-only LC
configurations offers the last-ditch hope to reverse the losses to low-cost
Tandy and IBM computers in the schools.
Preventing K-12 students from going DOS, in turn, would keep the university
campus as fertile territory for the Mac, particularly the Classic. If the
LC is the Mac's future, the Classic is a last nod to its past. The new machine
is what the Macintosh 128K should have been: a perfect computer for college
students. And college kids aren't likely to complain about the Classic's
limitations.
The lower-priced configuration, combined with the Mac's "hip"
image among the young adult set, will help Apple solidify the Mac's position
in its university strongholds. The $1,200 street price for the hard disk
configuration also should make it a popular yuppie home computer.
Business wants expandability. But the Classic is an instant orphan at major
businesses sites: it has minimal expandability, no color capability, no
virtual-memory support and no horsepower.
Instead, it's the Mac IIsi that offers Apple's first reasonably priced color
business computer. As a re-implementation of the Mac II concept, it reduces
six slots to one (although you must buy a slot adapter to use current NuBus
cards), while eliminating the need for a slot for 90 percent of its owners
with built-in 256-color video. Its most severe deficiency shows up during
heavy spreadsheet work, since it has no built- in numeric coprocessor. For
that, you must buy a $249 (list) slot adapter to get the 68882 floating-point
chip.
Cost still a problem. Unfortunately, the new machines do not eliminate all
of Apple's cost disadvantages. The current product lineup fragments Mac
expandability across NuBus and three types of direct slots, assuring that
production runs will remain shorter and board prices higher than in the
DOS community.
And both the IIsi and LC call attention to the excessive price of the SE/30.
Once the LC begins shipping, the SE/30 will be $1,000 more than a color
LC, with 80 percent of its horsepower, and about the same price as a comparable
color IIsi.
Consider CD-ROM for internal publishing (10/23/1990)
Joel West, "Consider CD-ROM for internal publishing," MacWEEK,
v4, n36 (Oct 23, 1990):34.
If you frequently need to distribute data and software throughout your company,
it might be time to consider using CD-ROMs.
Surprised? Don't be. The technology offers two powerful advantages: high
capacity (the read-only medium can store at least 550 Mbytes of information)
and low cost when produced in quantity ($2 per disc). Commercial software
publishers already have begun to embrace CD distribution for precisely these
reasons, and there's no reason companies can't do the same for shipping
updates of large in-house applications. Other possibilities include product
catalogs, maintenance manuals, company directories and newsletters, and
even university course schedules.
Powerful and compact. The compactness and durability of the discs also provide
important logistical advantages, particularly over floppies, which tend
to be the medium most used when people can't send their data over phone
lines or LANs. And you can't easily retrieve data segmented across a stack
of floppies but instead must copy the data to a local hard disk. By contrast,
a CD-ROM can be used either for distribution (copying data at the receiving
end) or for on-line retrieval.
Since every Mac has a floppy drive and every user knows how to use it, when
do you decide to go the CD route for sending out information? Obviously,
when each user has to cope with too many copies.
But what about other high-capacity alternatives, notably 150-Mbyte TEAC
tape cartridges and 45-Mbyte SyQuest removable disk cartridges? It depends
on the size of your distribution list. A CD-ROM is always more cost-effective
when you need to send 50 copies or more.
In fact, the cost of using a CD-ROM for internal distribution is surprisingly
affordable: the setup cost is about $1,200, plus $2 per disc (minimum order
100). Turnaround is about a week. What's more, the discs can be shipped
directly from the pressing plant to company sites around the world.
True, CD-ROMs cannot be modified or updated once the disc is pressed. And
that makes preparation of the master disk (usually on a large-capacity hard
disk) extremely important. But there are some side benefits for distributing
data. If you ensure that the master is virus-free, the read- only medium
can never be infected. Nor can data be accidentally deleted.
Keep it organized. Sending such large amounts of information requires that
careful thought be given to its organization. This may be as simple as an
introductory HyperCard stack that explains each component on the disc, launching
an application or database from a button in the corresponding description
card.
For other discs, it will make more sense to rethink your entire data- access
strategy. You may want, for example, to convert a series of static text
documents into a single interactive hypertext document.
Once the master is prepared, you must decide between three possible CD-ROM
formats that can be used with a Macintosh. If the disc will be read only
by Macs, then the Macintosh's native hierarchical file system is the easiest
and most natural format.
Two other formats -- ISO 9660 and the earlier High Sierra -- allow the disc
to be read on both Macs and IBM PCs and compatibles. For such a dual-platform
strategy to be effective, you also must have software for both the Mac and
PC that can make sense of that data, whether that retrieval software is
a commercial application or something developed internally. You also must
live with the user-visible file naming and organization limitations imposed
by the ISO and High Sierra formats, which force a name such as "Teacher's
Directory" to be changed to the uglier "TEACHERS.DRC;1".
The houses that do CD-ROM duplication, including Du Pont, Sony and 3M, can
help you prepare your master disk to meet the limitations of your chosen
format. They also can offer technical advice on becoming an effective CD-ROM
publisher.
Reading the fine print in Apple-Adobe truce (10/9/1990)
Joel West, "Reading the fine print in Apple-Adobe truce,"
MacWEEK, v4, n34 (Oct 9, 1990):38.
Despite the recent rapprochement between Apple and Adobe, the two companies
still have conflicting goals and priorities. Apple's goals include promoting
System 7.0 and its TrueType font standard, ensuring that every Mac owner
can have high-quality typefaces on the screen and on a printer.
Adobe's commitment (if any) to TrueType remains unclear, but its position
on the Mac is not: Like most software developers today, Adobe is dividing
its time among the Mac, Windows and other platforms.
The truce between the two companies may have been helped by Jean-ouis Gassee's
departure, but it also reflects Adobe's growing realization that it, rather
than Apple, must promote its goals in the Mac marketplace.
As always, Adobe wants to sell more printers containing PostScript interpreters.
But the company also is promoting its Adobe Type Manager, application packages,
and Type 1 fonts that work with both ATM and PostScript printers.
Focus on tomorrow. For users, Adobe's new emphasis on promoting its own
goals will become most evident with PostScript Level 2, due to ship with
printers released next spring. The company is developing its own Chooser-selectable
PostScript driver that will support Level 2 as well as the original PostScript
printers and provide better printing efficiency than the LaserWriter driver
that Apple ships with system software.
Like any 5-year-old piece of software, the Apple driver is overdue for a
complete rewrite. But that rewrite has been tied to the New Print Architecture,
due sometime after System 7.0. You won't see full PostScript Level 2 support
from Apple before then, so Adobe needs its own driver to take advantage
of Level 2 features.
Features for business. One Level 2 feature that is likely to be appreciated
by many Mac users is compression of bit-mapped images, both color and black
and white. A Level 2 driver and printer should speed the painfully slow
print time for scanned images, faxes and even HyperCard pages.
Level 2 also reflects a tacit recognition that, despite its roots as a typesetting
language, most of the installed PostScript printers are being used as general
business printers. The new language codifies PostScript extensions for additional
features, such as dual paper trays and two-sided printing, that are more
appropriate to form letters than page proofing.
Apple's LaserWriter driver never offered such features because Apple doesn't
sell such a printer product. Adobe's OEM customers do.
The HP factor. Despite its high profile, Adobe is not the only company with
a major role in Apple's printing strategy. A third major player, Hewlett-Packard,
is more closely aligned with Apple than is Adobe -- at least when it comes
to software standards.
Despite HP's decision to offer PCL Level 5 (the language of the popular
LaserJet III) as a competitor to PostScript in the IBM PC world, HP recognizes
that the language is irrelevant to Macintosh owners. In fact, HP has hedged
its bets by offering both PCL and PostScript.
When it comes to Macintosh standards, HP and other hardware manufacturers
will follow Apple's lead in a way that Adobe never will. Unlike Adobe, these
manufacturers have no stake in TrueType vs. Type 1 or QuickDraw vs. PostScript,
as long as major opportunities remain to sell third-party printers to Macintosh
users.
The products from such third-party printer manufacturers will become increasingly
important to Apple as it refocuses on its CPU business. With the delays
that Apple has had in developing its own 300-dpi plain-paper color printer
(in which Gassee took a personal interest), it is conceivable that Apple
will leave the color printer market to the third parties.
Five years ago, desktop publishing on the Mac meant using an Apple driver
with an Apple printer. Five years from now, neither one may be important
to the average user.
New salvos on the home-computing front (9/25/1990)
Joel West, "New salvos on the home-computing front," MacWEEK,
v4, n32 (Sept 25, 1990):34.
When Apple introduces its three new Mac models next month, the machines
will play an important role in the home computer market.
Once thought to be the province of computerized recipes and video games,
the home market has turned out to have many of the same needs and demands
as the business market. That's because many people are, in fact, using their
home machines to do office work at home.
The home computer market never met industry expectations five years ago,
but it gradually has grown to reach 4 million personal computers a year,
by one Dataquest estimate. It's big enough that both IBM and Tandy are mounting
major efforts to sell home personal computers, emphasizing two-income households.
Apple once owned the home market with the Apple II. But that product's failure
as a business computer contributed to its current poor showing in the home
market, since most professionals insist on home computers that are compatible
with their machines at work.
Today, the home market is too big for Apple to ignore. It's essential that
the Mac maintain market share if it is to continue to attract new hardware
and software from third parties.
The home machine. If advance reports are accurate, then all three new Macs
-- ranging from the entry-level Classic to the higher-performance IIsi --
should appeal to home users. But it's the Macintosh LC, Apple's first color
68020-based machine for about $2,500, that will have the biggest effect.
It will be the first Mac to provide color within the home price range. As
Apple has discovered, few individuals who have to dip into their own wallets
will pass up a $2,000 80286 color system to buy a 68000-based black-and-white
SE.
Ironically, the Mac revolution began in the home, with a few fanatics ignoring
industry trends and buying this untried machine. As with other home computers,
their buying motivation tended to fall in one of three categories of home-based
work: bringing work home, moonlighting and self- employment.
Bringing work home is the most common use I've seen, particularly with the
rise of telecommuting. For example, a city bureaucrat friend cranks out
manuals and brochures one day a week on his accelerated SE.
Many people I know who bought a Mac for their job also talk about "starting
a business on the side." Here, user motivation is much stronger than
with just a hobby. The father of a friend is, late in life, reaching computer
literacy by replacing his electric typewriter with a IIci. Although he will
use it to watch his business interests, he chose a Mac because it was best
for publishing his boating-squadron newsletter.
The home computer is even more important to those who make their living
at home. Another friend does graphic design on a Mac II in a spare bedroom.
Unfortunately, Apple's previous pricing strategy has hurt its efforts to
sell Macs for home use. Less-knowledgeable and less-affluent workers will
buy what they have at the office or what friends recommend based on their
own experience or simply on price. Apple is a loser on all three fronts,
unless it decides to license its older ROMs to other manufacturers to broaden
the Mac market.
The home program. Of course, software is also important. The original Jobs
appliance came with a word processor when software packages cost $125. Now
they cost $295. "Borrowing" software for the home is rampant.
IBM recognized the needs of the home market and now bundles software with
its PS/1. If Apple truly wants a slice of the lucrative home-market pie,
it, along with software publishers and dealers, must ensure basic capabilities
with every home Mac that Apple sells.
The package that will allow me to write letters, draw maps, calculate my
taxes and publish a newsletter -- all for $195 -- will be the one that dealers
sell with every home Mac. Now that Apple is finally getting off its duff,
let's hope software publishers respond in time.
Color accuracy: Still a long way to go (9/11/1990)
Joel West, "Color accuracy: still a long way to go," MacWEEK,
v4, n30 (Sept 11, 1990):40.
Two years ago, I bemoaned the lack of color consistency among monitors for
the Mac and between monitor and printed output.
The need for calibrated color is even greater today, since the Macintosh
is widely used for color applications (notably pre-press), previously the
sole province of high-end workstations. Average users also experience the
problem as colors shift when an image goes from being scanned to being displayed
and printed.
There's still no such solution in sight, even with next year's System 7.0.
Still, a number of third-party products have appeared, but they only partially
solve the problem.
For example, the Pantone Matching System, supported by many applications,
improves spot color but has no effect on scanned images or four-color printing.
TekColor from Tektronix Inc. (and SuperMac Technology's SuperMatch, which
incorporates TekColor) attempt to calibrate printed output to match colors
seen on the monitor. RasterOps, Radius and SuperMac all offer competing
(and incompatible) products that incorporate an electronic color sensor
in an attempt to compensate for color differences among devices.
But a complete solution would require Apple to define QuickDraw colors in
terms of scientifically measurable color values, as Adobe Systems Inc. has
done in PostScript Level 2, due next year.
Color on the Mac is defined as a percentage of device-specific RGB (red,
green, blue) values as produced by monitors. But such relative RGB colors
will vary among monitors or print differently among printers. An absolute
definition of what any given Mac-produced color is supposed to look like
would allow a monitor or printer manufacturer to calibrate the device to
a known standard.
But calibration of corresponding colors is only half the battle for color
accuracy. The other half is representing colors on a device within its gamut
or range of available colors. A color available on one device (say bright
red on a monitor) may not be available on another device, such as a printer.
One way to handle this problem is to clip the colors to fit the printer
gamut, mapping two shades of bright red that differ on the screen to the
closest red available on the printer. The other way is to compress an image's
color range to fit within the printer gamut, so that two on- screen shades
still will differ noticeably on the printer. The choice of the "best"
mapping approach will vary by color, application and personal preference.
Should these options be under user control (producing yet another "dialog
from hell") to benefit the professional designer? Or should the Mac
automatically "do the right thing" to simplify life for the average
user? The best approach would seem to be placing these choices under the
control of the application, so that they can be integrated into the image
creation and printing process.
PostScript Level 2, with its more precise definition of color values, certainly
will beat QuickDraw to the color-calibration punch. It does not provide
application control over these mapping options, which seems a serious limitation.
Nor will Level 2 allow user feedback.
The overall PostScript standard is intended to be device-independent, so
that a color image prepared on one device can be printed on another. But
the Mac's WYSIWYG precept implies that the user should know how a color
image will be printed as it is being created. In many cases, users will
want to limit their color selections to those available on the printer,
an interface that is implemented well by the TekColor Picker.
While some users can wait for Apple to define new interfaces and standards,
others need accurate color now. For them, a third-party product could be
a good interim solution. And when (or if) an Apple- defined Mac standard
does become available, the Mac vendors that offer solutions for accurate
color should provide upgrade paths for their products.
But if such an upgrade path is important to you, then, as they say on TV,
"get it in writing."
The shape of in-house development (8/14/1990)
Joel West, "The shape of in-house development," MacWEEK,
v4, n28 (August 14, 1990):30.
With its third birthday behind it and Version 2.0 completed, HyperCard seems
better-suited than ever for solving many of the problems that come up in
in-house development.
Almost from the beginning, HyperCard has been an excellent choice for one-of-a-kind,
quick development. And while simple applications can be built easily with
HyperCard, its nearly unlimited extensibility via HyperTalk and external
commands (XCMDs) allows developers to create sophisticated applications
as well. If those applications have been well- designed, they can be used
by any Macintosh user regardless of computer savvy.
With Version 2.0, HyperCard loses some of its un-Maclike rough edges and
becomes easier to use to author new stacks. HyperCard should be an important
part of the grab bag of tools for in-house development.
That's a surprisingly important role when you consider that three years
ago no one fully comprehended what Bill Atkinson's HyperCard was all about.
Some people referred to it as a database, while others emphasized its hyper-media
possibilities.
I prefer to think of HyperCard as a programming environment based around
a card metaphor, with built-in database, hypermedia and bit-mapped graphics
capabilities. And that means that you must carefully consider how HyperCard
applies to the task at hand.
For example, HyperCard can be a real pain to use for many database purposes.
A simple flat database is a much better platform for setting up routine
mailing lists, standardized data entries and reports.
But when you want more flexibility or customization, HyperCard shines. It
can, for instance, automatically complete fields based on a table look-up
or export data in an arbitrary format.
The new HyperCard allows this flexibility to be expressed in a fashion that
is more Maclike.
You're finally freed from the tyranny of one 342-by-512-pixel card; you
can design stacks around any fixed-size format that you desire. Version
2.0 allows you to resize windows and scroll within a card, although this
is more like Atkinson's MacPaint 1.0 than standard Macintosh applications.
It's also possible to have multiple stacks open at a time.
HyperCard 2.0 also supports several other features that make it easier to
develop stacks, including improved script editing and debugging, built-in
icon editing and floating tool palettes.
Such improvements cement HyperCard's advantages for in-house development.
It was already much easier to write and debug a HyperCard stack than a Mac
program written in C or Pascal. More people can become good HyperTalk programmers
than will ever be C toolbox wizards.
And while the resulting program may not have the polish of a custom C-based
application, it's a lot easier to write and maintain -- an acceptable trade-off
for a program that will be used by 100 people instead of 100,000.
All of these advantages also apply to the major HyperCard alternative, SuperCard
1.5. While HyperCard 2.0 nullifies many of SuperCard's authoring and debugging
advantages, SuperCard offers graphics advantages that are essential for
many applications. SuperCard provides an integrated gray-scale/color capability
and also supports MacDraw-type object graphics instead of HyperCard's MacPaint-style
bit maps.
In the future, both products' most important application will be as user-customizable
interfaces for information retrieval. Whether it's browsing through a commercial
CD-ROM or querying the company mainframe, this is an area where HyperCard
(or SuperCard) should be considered the tool of preference.
What's more, HyperCard will assume a greater role once applications are
available that take advantage of Interapplication Communications provided
by System 7.0. When that happens, HyperCard will become the front-end shell
that the Macintosh always needed but never had.
System 7.0 may leave average user behind (7/31/1990)
Joel West, "System 7.0 may leave average user behind," MacWEEK,
v4, n26 (July 31, 1990):52.
As everyone knows by now, next year's System 7.0 will offer more of everything.
But more is not always better. Despite Apple's pioneering efforts in ease
of use, System 7.0's power for sophisticated users adds a level of complexity
that leaves other users behind.
Don't get me wrong; it's still easier to set up a FileMaker database than
one in dBASE IV, or a LocalTalk network instead of NetWare. But when it
comes to ease of use, the Mac isn't what it used to be. And for that matter,
Word 4.0 isn't MacWrite 4.1, SuperPaint 2.0 isn't MacPaint 1.0, and Finder
7.0 isn't Finder 1.1g.
Gone is the simple elegance of the original Macintosh 128K. Now, Apple is
intent on positioning System 7.0 to compete with Windows 3.0 and OS/2. The
revised System, Finder and associated utilities are replete with features
that will appeal to hard-core enthusiasts.
Subscribing to and publishing Interapplication Communications editions may
be snazzy, but how relevant is it to the average user? Many of these features
will increase users' dependence on their local "Mac guru."
The average user's eroding self-sufficiency also can be attributed to the
Mac's inevitable shift from cult machine to just another business computer.
The early adopters were all Mac fanatics, but Apple has worked hard to expand
its market beyond fanatics. Many people no longer use the Mac because they
love it, but because it's part of their job. The Mac requires training,
support and administration -- just less of it than its competitors.
Most sites require their own system administrator, whether officially or
by default. Hardware fails, network connections are accidentally severed,
or maybe it's just time to upgrade all 27 CPUs to System 6.0.5.3.
The number of Mac users who can adequately tune the system heap is comparable
to the percentage of DOS users who can write an AUTOEXEC.BAT file.
Even when things are installed and running smoothly, the Mac is a far more
complex machine to run than when it was introduced, the result of a bad
case of creeping featurism.
A typical problem for novices has been the increased complexity of the Open
and Save As commands, which have little to do with the more elegant and
understandable folder meta-phor. Few novices I've watched grasp the hierarchical
structure of their hard disk. Instead, they leave documents in the application
folder, because that's the folder they get by default.
A recent Macworld magazine survey of its readers established MultiFinder
as a tool for the privileged few. Of those surveyed, 60 percent said they
used used it seldom or never, 13 percent used MultiFinder sometimes, and
27 percent used it most or all of the time. Anyone who thinks it's just
a lack of RAM has never watched a novice fumble through multiple applications
on a 9-inch screen.
But complexity is not a problem only for novices. I once felt a mastery
of my Mac until I discovered hard disks with non-viscous lubricating grease,
head crashes, low-cost power supplies and SCSI timing problems.
And today, faced with a mysterious software crash, my instinctive reaction
(justified or not) is a trial-and-error process of turning off INITs until
the problem goes away and, I hope, never reappears.
System 7.0 does offer some improvements in ease of use. Bubble help will
give more clues to the average user. A pop-up menu will become a more obviously
clickable item, and that will make the Open and Save As commands easier
to use. The Font/DA Mover has earned its retirement. And Laser Prep wars
will soon become a nightmare of the past.
But some of System 7.0's improvements could prove to be a handicap to new
users. For example, Apple-menu aliases can be set up to make it foolproof
for a novice to access favorite applications and often-needed documents.
But there's the danger that such users would come to depend on those preconfigured
options, unable to comprehend and use someone else's differently configured
Mac.
When it comes to empowering the individual, today's Macintosh is no Knowledge
Navigator -- or even a Macintosh 128K.
At last: a way to test Mac software (6/26/1990)
Joel West, "At last: a way to test Mac software," MacWEEK,
v4, n24 (June 26, 1990):24.
Compared with other computers, the Macintosh has always been a software-testing
nightmare. That's because the elements that make Macintosh applications
unique also make those applications extremely difficult to test. The Mac,
for example, has no command-script interpreter that knows about things such
as windows and menus.
Contrast that with Unix, which can test an application using a directory
of data files and command scripts that automatically put the software through
its paces.
To appreciate the problem, imagine developing a new graphics application.
You would want to test the application by creating a new document, modifying
it, saving it and printing it. Now imagine doing these tests with several
combinations of CPUs and peripherals, including at least a Plus and a Mac
II-family computer (that is, with and without a 68881) and with and without
color QuickDraw. For peripherals you would want to check out black-and-white,
256-color and 24-bit color screens. You also would test printing on a variety
of printer types and with an ever- smaller application memory size.
But there's never been an automatic way to repeatedly test an application
for multiple revisions and systems. In the early Mac days, Apple offered
a crude mechanism called the "monkey," which generated seemingly
random but repeatable events. However, this undocumented tool was about
as likely to emulate a power user as a real monkey would be to type the
works of Shakespeare.
Apple next offered MacroMaker, which provided some capabilities for repeatable
events. But MacroMaker is oriented more toward testing a specific mouse
position than what the user can see at that position. And MacroMaker does
not allow you to execute macros based on what appears on the screen or to
operate on a series of windows.
But last month Apple introduced a new utility at its Worldwide Developers
Conference that will finally make the job of testing Macintosh applications
much easier.
Called Virtual User, this new set of Macintosh Programmer's Workshop- based
tools will generate test scripts. It runs on an AppleTalk network, allowing
a single CPU to control any CPU running the VU Agent Startup document (INIT).
A key feature is that scripts are written as ordinary text files in a general-purpose
language that looks like a cross between Pascal and Smalltalk. Virtual User
scripts can operate on a list of windows or dialog boxes, although they
are particularly good for testing all the options within a dialog box.
Scripts can make decisions based on a partial match of the expected and
actual result. For example, a script can operate on any window with a title
whose first word is "Untitled."
Somewhat like MacroMaker, the VU Recorder application can be used to record
scripts by watching what you do. Those recorded scripts then can be played
back in combination with handwritten scripts.
The current design of Virtual User does have some inherent limitations,
however. It's possible to operate symbolically on the known application-independent
toolbox elements, but not on the application- specific elements of a window.
For example, the Circle tool of MacDraw can be selected only by position
-- 75 pixels down from the window corner -- not by its appearance. There
also is no way to say, "Select the top- right corner of the front-most
polygon and drag it up 45 pixels."
And developers wanting to test the international version of a program will
have to translate all the menu and dialog strings within the VU script.
Obviously, Virtual User will be more flexible and powerful with AppleScript,
Apple's planned application-control language. But AppleScript is not likely
to ship until next year, with AppleScript- compatible applications lagging
even further behind.
Still, Virtual User is a great step forward for Macintosh application development.
And when it is divorced from MPW and made friendlier to users, it will become
a mainstay for in-house training and development.
Quenching the quest for quicker QuickDraw (5/8/1990)
Joel West, "Quenching the quest for quicker QuickDraw," MacWEEK,
v4, n18 (May 8, 1990):24.
One way to make a Macintosh faster, and one that is now rapidly gaining
attention, is graphics acceleration.
This trend is highlighted by the release of Apple's first graphics accelerator,
the Macintosh Display Card 8*24 GC, one of several products now available
that provide faster video-display drawing for Mac II-family computers.
Graphics acceleration is but one part of more-powerful and complex system
design that uses separate processors for separate functions. On an inexpensive
IBM PC or compatible, the main CPU performs all functions, including general
control, math, I/O and video-display updating. On more advanced computers,
including a Mac IIfx with the 8*24 GC, separate processors handle each function.
For the Macintosh, graphics acceleration for off-the-shelf applications
means accelerating QuickDraw, the standard interface for screen display.
When is such acceleration necessary? For most users, an SE/30 with a black-and-white,
342-by-512-pixel screen is plenty fast. But with true color, you have denser
data: 24 times as much data per pixel and thus 24 times as much data to
be imaged and displayed in the same amount of time.
From a "real estate" standpoint, a two-page black-and-white display
has six times as many pixels as the SE/30. A 19- or 21-inch, 24-bit display
will require more than 100 times as much data as that one-bit SE/30. And
that means that waiting for QuickDraw to graphically update the screen can
be painfully slow.
To satisfy the demand for better graphics performance, Radius Inc., RasterOps
Corp. and SuperMac Technology offer QuickDraw accelerators to drive 19-inch,
24-bit color displays as well as Apple's 13-inch color monitor.
Radius and RasterOps require a separate video board, while the SuperMac
and Apple accelerators are built into a video board. Apple's video board
is designed for use only with its monitors, which include 24- bit color
on its 13-inch display or eight-bit gray-scale on its 21-inch two-page display.
With a 30-MHz RISC chip, Apple claims the 8*24 GC has more horsepower than
even the 40-MHz 68030 used in its IIfx.
The accelerated video boards work as graphics coprocessors, much as a Motorola
68882 works as a floating-point coprocessor to the 68030 CPU. Individual
operations are passed from the CPU to the accelerator, and then the CPU
continues with the next task. The accelerator generally won't take a second
drawing request until the first one is complete. All perform QuickDraw calls
on the coprocessor instead of on the main CPU, allowing the boards to work
with current applications.
A limitation of all but Apple's board is that they accelerate only on-screen
graphics drawing. Many applications, such as MacDraw II and HyperCard, prepare
off-screen bit images of their on-screen display, which are then copied
on-screen for faster updating on non-accelerated video displays. Accelerators
that speed up only on-screen drawing will have little effect on such programs.
Applications that use Apple's new recommended off-screen drawing approach
will be accelerated by both Apple's and RasterOps' boards.
Radius' QuickCAD accelerator board provides a different approach. Instead
of accelerating QuickDraw, QuickCAD requires that programmers use a Radius-supplied
Startup document (INIT) that implements vector-oriented display lists. This
accelerates only niche programs, such as CAD, that rely on display lists
and that are rewritten to support QuickCAD.
Besides price, a major factor in deciding which (or whether) to buy is whether
the board works with the monitors you own.
Comparative performance is a bit trickier, since the results will differ
by application and even by document. Given the dozens of QuickDraw calls,
the designer of a QuickDraw accelerator must decide which calls are most
important to accelerate and how much effort to put into each. For example,
scrolling 24-bit images might be a more significant bottleneck than filling
an arbitrary polygon.
OOPS: waiting for the other shoe to drop (4/17/1990)
Joel West, "OOPS: waiting for the other shoe to drop," MacWEEK,
v4, n15 (April 17, 1990):34.
In 1988, Apple's most famous ex-employee (Badge No. 0) threw down the gauntlet
for Apple and its approach to software development.
Steve Jobs claimed that his new development environment, NextStep, would
simplify software development the way that the Macintosh had simplified
using a computer. Easier software development presumably means having better
software (both commercial and in-house) sooner.
After NeXT's announcement, Apple hinted that its own far-more- impressive
tools would be forthcoming to meet the challenge.
But it's been 18 months of waiting for the other shoe to drop. And in that
time nothing from Apple has made an appreciable difference in the overall
ease of developing Mac software.
Next month's Apple Worldwide Developers' Conference would be a likely venue
for any major announcements. But so far, all that's been seen publicly have
been small incremental enhancements, accompanied by periodic exhortations
and hype.
In the meantime, what we have is ever-larger Macintosh applications from
ever-larger programming teams with ever-longer delivery dates.
Apple has long been aware of the problem. As I've noted in earlier columns,
Apple was an early innovator in object-oriented programming systems. OOPS
is a more-effective means of structuring complex software applications.
When coupled with a library of standard objects, development time is reduced
significantly.
Apple, for example, developed one of the first Smalltalk interpreters, which
was later killed in favor of two third-party products. Its MacApp product
is one of the most extensive and practical object libraries, and commercial
products written with it have been shipping for the past three years.
HyperTalk and AppleScript are first cuts at an object-oriented replacement
for BASIC. And, with its new Finder in System 7.0, Apple will be one of
the first computer companies to release core system software written in
C++.
This track record notwithstanding, not one leading application has been
written using Apple's OOPS tools. And in the time since Jobs' challenge,
Apple has offered nothing that dramatically simplifies Mac- application
development.
The only news I've seen recently from Apple was an attempt to (pardon the
marketing jargon) reposition and differentiate Apple's efforts by modifying
the buzzwords. Instead of the standard industry acronym "OOPS,"
Apple developers should henceforth refer to Apple's unique suite of offerings
as "object programming."
Of course, the annual industry technical conference will still be called
OOPSLA, the rest of the industry will still be OOPing it up, and Apple'
product offerings will remain unchanged. But we can all rest happy that
there's a new name for it, right?
Meanwhile, it's been third parties making the small breakthroughs in ease
of development. The new Think Pascal development environment offers both
MacApp and a reasonable turnaround. Steve Jasik's Incremental Build System
speeds up ordinary Macintosh Programmer's Workshop development, although
many developers find it cumbersome.
And Apple? It still has nothing like NeXT's Interface Builder, an integrated
front end for prototyping applications.
More significantly, Apple has lost its "leading edge" mindshare
to NeXT. And that can't be regained through "repositioning," but
by innovative product breakthroughs.
The most important problem facing developers in the 1990s is that microcomputer
software, no longer the toy programs of the early years, is now approaching
the size and complexity of mainframe software.
More lines of code mean more work to develop that code, to test it and to
maintain it. Any method that does not reduce source-code size is, at best,
a short-term solution.
Apple has long employed some of the leading OOPS visionaries. How that vision
will provide new Macintosh applications is yet to be seen.
Here's hoping Apple's new pirates are at work
(4/10/1990)
Joel West, "Here's hoping Apple's new pirates are at work,"
MacWEEK, v4, n14 (April 10, 1990):44.
Here's hoping Apple's new pirates are at work
Two unrelated news items prompted me to contemplate the role of technical
leadership in Apple's future.
One event was Apple's decision to reorganize product development yet again,
this time in response to the departure of Jean-Louis Gassee. The second
was the Chapter 11 filing by Kaypro Corp., once a highflier in the microcomputer
business.
What does the latter have to do with the Mac? Not much. But the failure
of Kaypro is yet another (albeit little-noticed) object lesson for industry
companies, including Apple.
Kaypro's strongest growth, ironically, came about from the death of the
company it set out to emulate. The 1983 bankruptcy filing of Osborne Computer
left Kaypro alone in the low-cost transportable computer market.
But Kaypro stuck with CP/M while the rest of the world went with MS- DOS.
Once customers decided that Kaypro's transportables were obsolete, no amount
of promotion made any difference.
Instead, Kaypro ended up late to the IBM PC-clone market, offering products
equivalent to those offered by everyone else.
Now look at the Mac, which uses a basic architecture that is 6 years old.
Its market share seems to be peaking. The problem is that unique is not
enough, because unique is not forever. People stopped buying Kaypro's products
and started buying Compaq's because they could get a transportable that
was compatible with the other micros in their office.
Today, people buy PC compatibles for their home because, for less than $1,000,
they can buy a color machine that works with the software in their office.
They buy PC clones as portables because they have a wide range of choices,
from 1 pound to 15 pounds, from $1,000 to $5,000. And again, the clones
work with the other computers in the office.
Many of the Macintosh peripherals vendors find themselves competing with
Apple in monitors, hard disks and so on. So they are broadening out to offer
PC-expansion products, since that is what works with computers in the typical
office.
And almost every Macintosh software company is turning its attention to
Windows and OS/2 because, again, it runs on the most popular office computers.
For the true believers and Apple shareholders, the latter is the most disturbing
news. Despite starting from scratch against the seemingly invincible IBM,
Steve Jobs, Guy Kawasaki and others at Apple were able to convince many
developers that the Macintosh was so exciting that it demanded their attention.
Apparently that is no longer true.
Despite the tinkering with sales and the blame heaped upon Allan Loren,
all the marketing in the world won't make up for not having products that
the market wants. Just ask Kaypro.
With Apple's 18-month to two-year lead time on new products, it's too late
now to start work on products that will be seen in 1990 or early 1991. For
the next year, Apple must continue on the path Gassee seemed to set, improving
execution and building self-confidence. Apple marketing must price and promote
its aging product line aggressively.
The decision by John Sculley to run R&D is, I hope, merely an interim solution
rather than an ego trip. More than anything else, Apple's R&D
efforts need a new direction, new energy and new blood.
Sculley's taking direct charge could help that. Perhaps the managers reporting
to Sculley -- Ed Birss, Don Casey and Larry Tesler -- will have greater
visibility and resources now that their boss is chairman of the board.
But Apple can't forget that nothing in this industry lasts forever. CP/M
died, the Apple II is giving up the ghost and MS-DOS is beginning to creak.
The Macintosh will die too, and that day is not as far off as it might seem.
Nearly a decade ago, Apple mobilized its advanced research into a pirate
development team whose work helped save the company.
Sculley has built the company's mission on the vow to be different. For
his sake and Apple's, I hope the next generation of pirates is already hard
at work. n
Ins and outs of buying a Mac fax modem (3/20/1990)
Joel West, "Ins and outs of buying a Mac fax modem," MacWEEK,
v4, n11 (March 20, 1990):34.
A fax modem is to the fax machine what a printer is to the copier. Instead
of scanning in an image from an existing piece of paper, as is done on a
facsimile machine, a fax modem generates the image electronically.
A fax modem also can be used to receive faxes electronically rather than
directly printing them to paper. The fax modem exchanges a compressed bit-mapped
image across phone lines with an ordinary office fax machine.
Fax modems as internal expansion boards have existed on IBM PCs and compatibles
for years. Although the graphics required for a fax modem would appear to
be a natural match to the Macintosh, Mac products have yet to live up to
their potential.
Part of the difficulty is particular to the Mac. For example, users expect
a product to be easy to use and well-integrated with the Mac operating system,
and those requirements demand more complex software than on a PC. Mac products
are also more expensive than those for the PC, because products have to
include an external case and power supply to run across the entire Mac line.
Beyond that, the proper role of both Mac fax modems and PC fax boards is
yet to be realized.
For example, the superior quality of fax-modem output makes it better for
sending computer-generated images. Fax modems also have a marked advantage,
as they support the ability to send the same message to multiple recipients.
The best fax software will even allow you to keep your list updated from
your existing Mac-based database, something you'll never get on a stand-alone
fax machine.
But a fax modem's advantages are less dramatic at the receiving end, in
that it forces you to leave an expensive computer switched on to receive
messages. And as with any high-resolution bit-mapped document, faxes are
slow to print on a PostScript printer.
A fax machine, on the other hand, automatically prints your messages as
they arrive -- something that background fax software can't do in today's
Mac printing architecture.
A fax modem does have an advantage in forwarding a received message to another
fax without the serious image degradation of a fax of a fax data transfer.
So how do you choose a fax modem? Your first concerns, of course, would
be the usual ones of price, availability, warranty and support. Generally,
the 9,600-bps products are more expensive than 4,800-bps models, but the
price difference can be recovered if you send a lot of long-distance faxes.
But fax modems can be difficult to compare, because a good product includes
a complex mixture of hardware, firmware (software in the modem) and Mac-based
software. Software includes a Chooser-selectable output driver; low-level
drivers for sending and receiving data; and a separate application that
views and prints incoming faxes and routes complicated, multiple-document
outgoing jobs.
If your fax modem will be a stand-alone device, you want a 1,200-or 2,400-bps
model or some way to switch between the fax modem and your existing data
modem.
For most offices, a more important issue is networking, since each fax modem
requires a separate dedicated phone line. After all, you wouldn't want to
publish three or four incoming fax numbers for a 20- person office.
But the bottom line for most people is reliability and utility, areas that
have caused problems for first-generation Mac products.
The importance of software reliability should not be underestimated. The
Chooser driver may not be compatible with your application or the firmware
may not send correctly to a particular fax machine. Low-level drivers may
not be compatible with Startup documents (INITs) and networking software;
they could even bring your Mac to its knees when transmitting faxes.
As with any complex product, use a review to narrow your choices down to
two or three products, or look for a friend's recommendation. Better still,
find a product that you can try for a week, either on loan or with a money-back
guarantee, to see if it meets your particular needs. n
;Factoring System 7.0 into buying decisions (2/20/1990)
Joel West, "Factoring System 7.0 into buying decisions," MacWEEK,
v4, n7 (Feb 20, 1990):30.
With Apple's recent update of its announced plans for System 7.0, it's time
for many users to make some choices.
The publicly estimated shipping date -- summer 1990 -- reflects only about
a four-month slippage from estimates last year, so the current date can
be treated as realistic. One of the reasons that Apple can feel confident
that System 7.0 will ship this summer was that it dropped several features
that won't be ready in time, including the print architecture and the new
text-layout routines.
There are more than enough features left, however, for System 7.0 to have
a greater effect on users than any software change since the Macintosh was
released.
Many of these features will require application developers and hardware
vendors to begin modifying their products in the coming six months. Since
it can take three months to a year for these modified versions to be shipped,
the new versions will only begin to be available during the second half
of the year.
Among the most important changes will be the redesigned of applications
to take advantage of Interapplication Communications (IAC). Since it will
require significant work on developers' parts to work with IAC, Apple's
new breakthrough will have little effect until the new applications are
available. A much easier task for developers will be making applications
"32-bit clean" (able to run in Apple's new 4-Gbyte addressing
mode) and taking advantage of virtual memory.
Printer drivers (including Apple's) may need revisions to take advantage
of outline fonts, as may programs that contain advanced text capabilities,
such as page-layout applications.
All software should eventually be revised to take advantage of Apple's new
Help Manager, which will provide a standard interface for context-sensitive
help across all Macintosh software. But users will see an immediate benefit
with the on-line help provided in Apple's software, such as the Finder.
This new Finder -- the first full rewrite since it was released - will be
among the most visible improvements when System 7.0 ships. Also, System
7.0 will include Macintosh FileShare, the peer-to-peer file sharing software.
These new features will reduce or eliminate the demand for several third-party
products. The largest such group will comprise those that perform file sharing,
such as TOPS.
How should System 7.0 be factored into buying decisions in the next six
months?
There are few reasons to hold off on buying applications, video boards or
printers, since these types of software (or the software drivers for the
hardware) are likely to be fully upward-compatible with System 7.0. About
the only reason to wait would be to see if the software will support some
specific new feature.
In two areas, I would purchase only those products that take advantage of
new features. One area is the Communications Toolbox, which affects products
such as networked modems and serial I/O boards.
Another is software drivers for third-party hard disks. Although existing
drivers will work under System 7.0, those that are revised to take advantage
of asynchronous SCSI I/O and to be fully compatible with virtual memory
should be much faster than those that aren't. How can you tell which to
buy now? There's no way to know for sure, but I would expect that any leading
drive vendor that has made a major investment in software would continue
to do so.
Should you go out right now and buy a PMMU (paged-memory management unit),
the Motorola 68851 chip that allows a Mac II to take advantage of virtual
memory? I can't see why, because these have been getting cheaper since System
7.0 was announced and they're not much use without A/UX or virtual.
But if you don't already have the recommended 2 Mbytes of memory, buy RAM
as you can afford it. More memory is usable even before System 7.0 comes
out, and the prices have plummeted: I paid $72 per 1-Mbyte SIMM (single
in-line memory module) last month vs. $250 a year earlier from the same
dealer.
And with predictions of a weaker dollar, which would raise prices, and overcapacity
in semiconductor manufacturing plants, which would lower prices, it's hard
to predict where RAM prices will be six months from now.
Comm Toolbox: bonanza for big sites (12/12/1989)
Joel West, "Comm Toolbox: bonanza for big sites," MacWEEK,
v3, n44 (Dec 12, 1989):38.
The Macintosh Communications Toolbox will be a part of System 7.0 that remedies
several of the Mac's serial communications deficiencies and provides a modular
software architecture for a wide range of Apple and third-party products.
The Comm Toolbox will affect how developers write communications software.
The new standards also will open opportunities for developers of communications
hard-ware, such as modems or serial ports connected to LocalTalk networks,
or internal serial port expansion boards.
Perhaps most far-reaching, the Comm Toolbox may create a new class of applications
that are not primarily communications products but use communications as
part of their function.
This potential should be especially interesting to the in-house developer.
Instead of writing communications routines, you simply call existing routines
from the toolbox. Thus, you can concentrate on the specific capabilities
you need, such as mainframe data retrieval.
The Comm Toolbox may prove most useful for those sites connecting Macintoshes
to mainframes and minicomputers. Tie-ins to Digital Equipment Corp.-based
networks also are planned. And while the Comm Toolbox won't replace gateways
between LocalTalk and Ethernet networks serving Unix workstations, it could
be used to provide reliable mail-forwarding between Macintosh and Unix mail
systems.
Products that take advantage of the Comm Toolbox can be divided into three
categories. At the highest level, applications (such as a terminal program)
call the Comm Toolbox to perform communications operations. At the lowest
level, drivers allow the Comm Toolbox and compatible applications to use
specific hardware.
In the middle, three managers provide a standard user interface (used by
applications) for selecting plug-in "tools" in the System folder.
The Connection Manager lets you select a particular serial port, its transfer
rate, parity and so on. Apple supplies a modem tool (among others) that
handles such details as tone or rotary dial for a Hayes-compatible modem.
The Terminal Manager controls the emulation characteristics of the terminal
tools, such as the Local Echo and Screen Width settings.
Finally, the File Transfer Manager controls tools that implement serial
file transfer protocols, such as Xmodem or Kermit.
Additionally, Apple has provided a way for hardware developers of products
such as NuBus serial port cards or network modems to register their device-specific
drivers with the Comm Toolbox. This permits any Comm Toolbox-compatible
application to directly select hardware ports.
Such mix-and-match flexibility will be an important factor behind the power
of the Comm Toolbox once it is adopted and supported by the various communications
software developers.
For example, a vendor with an established proprietary communications protocol,
such as Communications Research Group with its MacBlast, could site license
a Macintosh file transfer tool with its mainframe communications software.
MIS directors could choose to combine that protocol with the terminal program
that best meets their users' needs.
Although the Comm Toolbox is available before System 7.0, its immediate
effect is uncertain. Will Apple promote a list of products compatible with
the Comm Toolbox? Will most developers offer Comm Toolbox versions of their
products now? Or will Apple, its developers and its customers hold off until
the Comm Toolbox ships with System 7.0?
Certainly Apple has not finished integrating the Comm Toolbox with the rest
of its software architecture. AppleTalk, the Printing Manager and MacWorkStation
should be extended to use serial ports via the Comm Toolbox. And the Comm
Toolbox does not yet provide a standard for the crucial problem of port
allocation.
But, unlike other areas of system software, there are no competing standards
announced or on the horizon. For any in-house developer, the time to evaluate
the Comm Toolbox is now.
Should you invest in Adobe Type Manager (11/21/1989)
Joel West, "Should you invest in Adobe Type Manager?" MacWEEK,
v3, n42 (Nov 21, 1989):36.
The recent introduction of Adobe Type Manager raises a number of key questions
for large Macintosh installations. For example, when will it be useful and
how will ATM fit into future font strategies?
The first question about ATM that must be answered, however, is, "What
is it?"
Adobe Type Manager provides scalable screen and printed text using outline
font definitions. It is based on the same font-rendering technology used
in Apple's LaserWriter printers but works more like the font software found
in many printer drivers.
ATM traps QuickDraw's standard text procedure and, for Adobe fonts, draws
those fonts using Adobe's font definitions. The $99 package includes four
font families: Times, Helvetica, Courier and Symbol. It uses at least 160
Kbytes of RAM, more if you set the font cache beyond the 64-Kbyte minimum.
ATM's main benefit is providing high-resolution fonts at any size. For example,
the bit-map fonts that come with the LaserWriter II SC print smoothly at
up to 24 points; ATM can print 60 or 61 points. And unlike QuickDraw's slanted
text, ATM provides true italics as well as more- distinct bold at large
sizes.
However, ATM's rotated text is not as good as on many printers or drivers.
The few applications that cleanly rotate existing bit-map fonts will look
good when printing with ATM, but those that print ugly rotated bit maps
will still look ugly with ATM, even if they look nice on the LaserWriter.
Nor can ATM improve font quality in painting programs or HyperCard.
There's more behind ATM, of course, than just font-rendering software. Its
introduction is a calculated effort by Adobe to influence typeface loyalties
before Apple delivers a competing outline font format with System 7.0.
Once you buy the ATM razor, Adobe hopes you will buy a lot of razor blades
in the form of Adobe typefaces. For $198, you get the next seven font families,
giving you the full LaserWriter IINTX complement, but each additional family
typically runs $185.
This can add up quickly because ATM and its fonts are licensed per CPU,
not per printer. A site that has invested heavily in Adobe fonts will be
less interested in buying fonts in Apple's System 7.0 format -- or so Adobe
hopes.
Against this you must weigh Apple's promises of several bundled typefaces,
faster on-screen display, and compatibility with most applications and hardware,
including PostScript printers.
Typographically, Apple's as-yet-unnamed font format is a decade more advanced
than Adobe's existing product. A font that fully exploits Apple's new format
will produce excellent results on 72-dpi screens as well as on 2,500-dpi
typesetters -- unlike Adobe, which has made subtle trade-offs at either
extreme. But Apple's fonts are proving extremely time-consuming to produce,
since a simple typeface conversion provides no such quality breakthrough.
Because it will come with every Mac, Apple's new format will become the
primary standard for most developers, no matter what Adobe does in the interim.
So deciding the long-term value of an Adobe font investment requires assumptions
about Adobe's future in the Mac market.
Many loyal desktop publishing users wonder why Apple and Microsoft are "picking
on" Adobe. When Adobe was the only game in town, it developed a reputation
for arrogance (as do many industry leaders) that encouraged companies to
find other alternatives. That's why Apple and Microsoft will be sharing
Microsoft's PostScript clone and Apple's fonts.
The release of ATM suggests a more hungry, less complacent Adobe, and perhaps
one that finally acknowledges what every other developer in the Macintosh
market knows: Only Apple makes the rules.
The old Adobe would bear a grudge against any company that used a PostScript
clone. Will a new Adobe work with Apple anyway, thus maximizing the role
of Adobe and its customers in the Mac market?
After all, if Microsoft can bury the hatchet with Apple, why not Adobe?
Made in the United States (11/7/1989)
Joel West, "Made in the United States," MacWEEK, v3,
n40 (Nov 7, 1989):38.
Recently, Apple made a decision that could have a crucial impact on its
future. Apple decided not to co-sponsor U.S. Memories Inc., a consortium
set up to manufacture dynamic RAM modules.
If it's right, Apple saved its cash for a rainy day. If it's wrong, users
may someday face a situation where Macintoshes are hard to come by. That's
because, as last year's DRAM debacle dramatically demonstrated, Apple is
critically dependent on the pricing and availability of outside components.
As with other large computer companies, Apple tries to arrange different
suppliers and to sign long-term contracts to assure a reliable supply of
those components. But some forces are larger than even a company the size
of Apple can control. That's why IBM, Digital Equipment Corp., Hewlett-Packard
and AT&T as well as four semiconductor companies co-ounded the manufacturing
consortium.
Apple reportedly declined to join because it had a better use for its money
and because it saw no customer benefit in joining.
Apple has manufacturing plants in California, Ireland and Singapore, with
components also coming from Japan, Hong Kong, Korea, the Philippines and
Mexico.
IBM, DEC and H-P have an equally international view but recognize that two
keys to Japanese success have been vertical integration to supply many components
and tight partnerships with outside suppliers.
Today, Apple buys most of its DRAMs from potential competitors: Toshiba,
a leading laptop maker, and NEC, the dominant computer company in Japan.
Cray Research used a similar approach until its Japanese supplier started
making supercomputers, then delayed supplies of key components to Cray.
Apple also has a close relationship with Sony, from which it buys floppy
disk drives, hard disks, CD-ROM drives and video tubes. Apple already buys
sound chips from Sony and would likely buy DRAMs if Sony made them.
But such a dependence on overseas suppliers exposes any company to political
and economic forces that may be beyond its control.
Apple and other companies can influence U.S. trade policy via extensive
lobbying. But there's little that any company can do about the price of
the yen or about international politics. Some Japanese politicians are even
thinking about firing the first salvo in the next trade war.
As last year demonstrated, scarce DRAMs both increase user prices and reduce
overall sales. Apple's sales growth is crucial to existing Macintosh sites
because that money is used to fund increasingly complex R&D.
Certainly, lack of funds was not behind the Apple decision, since the company
is sitting on three-quarters of a billion dollars. Apple could use that
money to assure a DRAM supply by acquiring a small, innovative semiconductor
company, such as key supplier VLSI Technology.
But since the days of the Lisa "Twiggy" drive, Apple has been
leery of designing its own components when it can buy them. If Apple continues
to reduce its manufacturing contribution, it could eventually become strictly
a design company.
Perhaps that is Apple's goal: to become a software company that merely licenses
its software and firmware to manufacturers.
In his new book, "The Japan That Can Say 'No,' " Sony Chairman
Ako Morita argues that such "hollowing out" is not in the United
States' long- term interests: "When people forget how to produce goods,
and that appears to be the case in America, they will not be able to supply
themselves even with their most basic needs."
Advice to the in-house Macintosh developer (10/17/1989)
Joel West, "Advice to the in-house Macintosh developer," MacWEEK,
v3, n37 (Oct 17, 1989):34.
As companies purchase more Macintoshes, their need for custom programs increases
as well. Such custom software may range from a few spreadsheet macros to
a full-blown application. In the latter case, the pitfalls facing the in-house
developer parallel those dealt with by commercial developers.
Usually, though, the goals of the in-house developer are more limited. The
software to be developed tends to have a narrow focus, and the target audience
is smaller and better defined. Commensurately, development resources also
are smaller.
For many in-house developers the best approach is to turn to the shortcuts
long favored by value-added resellers, creating custom applications with
HyperCard, SuperCard, 4th Dimension or Excel instead of the programming
tools typically used for developing million-selling applications.
But regardless of whether the software is intended for dealers' shelves
or for in-house use, the development pitfalls are the same. In each case
the final goal is an intangible product, leading to an amorphous definition
of what needs to be done and how much work it will take.
Mac software must be well-designed to realize the benefits expected of the
Mac. If the interface is not clear, coherent and consistent, you lose the
ease-of-use advantage that may have justified the Mac's purchase. At the
same time, if the software is not designed to share data with other programs,
you lose the interoperability that is the basis of applications such as
desktop publishing and desktop presentations.
Both general and Mac-specific development headaches can be mitigated in
the design phase by having a good spec or series of specs. Depending on
the project, you need to prepare one or more of the following:
>Requirements document: a laundry list of what the software must do,
but not how.
>User's spec: a plan that often can be handled by writing a draft user
manual before writing the software.
>Internal spec: a description of the implementation approach, which is
crucial when more than one programmer is involved.
You may need additional plans, schedules or specifications if exactly meeting
your schedule and budget is important. If you aren't concerned about project
overrun, it may be easier to run a looser, less structured development effort.
Even the best designs require validation, and that is particularly true
for the interface of a Mac program. User testing is essential to the acceptance
of any new package. Even if time is limited, you must test your assumptions
on at least one or two prototypical users.
But as with any other software project, the planning and resource allocation
must go beyond the initial development period to the entire software life
cycle.
Software doesn't rust or wear out. Instead, most problems stem from a bug
or from using the software for something it wasn't designed to do. The latter
can include requirements that could not have been anticipated, such as compatibility
with new applications or new versions of system software.
But many problems can be anticipated, such as compatibility with various
Macintosh CPUs. This might include testing on the entire Macintosh family
and checking to see if the results look good on black- and-white, eight-bit
color and 24-bit color displays. This also includes designing for any screen
size.
Why design for the largest possible target, even if you're sure the software
will be used by a much narrower audience? Well, suppose the top manager
brings in his or her personal machine from home or, when it comes time to
rent machines for a temporary overload, your standard configuration is not
available. You'll be forced to do a panic update on someone else's schedule
rather than your own.
Finally, learn as much as you can about Mac development. Books, magazines,
sample programs and development tools are available, and $30 spent today
pales compared with a half-day's development later on.
Along these lines, Apple has set up a program to provide technical information
to the in-house developer. If your budget can handle it, joining the Apple
Associates program may be the best place to begin your development effort.
n
Font wars on the computer front (10/13/1989)
Joel West, "Font wars on the computer front," MacWEEK,
v3, n35 (Oct 3, 1989):30.
The font solution under development for the Mac's System 7.0 is affecting
the shoot-out over the font approach for Microsoft's OS/2 Presentation Manager.
Together, these two solutions will bring digital typography to more people
than ever before. At the same time, setting the font technology for these
systems could make or break the companies that have developed such technologies.
Software for generating digital typefaces -- scalable outline fonts -- can
be evaluated using several criteria. Speed is the most noticeable, but also
important are the code size and memory usage.
Typefaces can differ dramatically in type quality at small point sizes because
of the "hints" used to make the scaled font more readable. The
ease of generating new typefaces also varies depending on the format.
A number of vendors have been actively licensing their intelligent typeface
technologies to microcomputer application vendors and printer manufacturers.
In addition to Adobe, the most visible is Bitstream Inc., the one company
that lives or dies by retail sales of microcomputer fonts. Its Fontware
is used by many Post-Script clone printers, such as the Qume CrystalPrint.
On the Macintosh side, Fontware algorithms are incorporated in the printer
software for GCC printers. To compete for OS/2, Bitstream has developed
a faster algorithm, code-named "Speedo," which recently shipped.
Another perspective is held by Compugraphic, a maker of stand-alone typesetting
systems. Its Intellifont engine has one major ally, Hewlett- Packard, which
uses it in the software for the DeskWriter, H-P's first Mac-only printer.
The third typeface supplier is URW, the German type foundry. URW's Nimbus
software is now shipped by Mirus and Palomar, while URW typefaces are also
used by a new engine, the Nimbus Q, from The Company, a URW affiliate; the
first Nimbus Q driver has been shipped by GDT Softworks. All of these formats
were taken into consideration in the design of Apple's font format, and
all have been players in the OS/2 font decision.
Apple's engine appears to be fast with modest memory consumption. It renders
well at small sizes; most engines target 25-pixel type, while Apple does
well at half that size.
By designing its own format, Apple has chosen an approach that provides
the most opportunities for third-party type suppliers. Although this will
be a less lucrative market than selling proprietary font engines, it is
open to all of the font engine vendors as well as type foundries.
But the fundamental differences between two proprietary typeface formats
can make it difficult to translate between formats.
Each format starts with digitized characters with several thousand pixels
in each dimension. These images are translated into a series of straight
lines and curves. The curves may be circular arcs, quadratic splines or
cubic splines. Apple's format is based on quadratic Bezier curves.
Although using the same type of curve makes it easier to move from one format
to another, ordinary number crunching can be used to refit the curves. Not
so with hints, which are fundamentally different. Apple has solved this
problem by defining a format that can represent all of the popular hinting
mechanisms.
This sets a difficult path for OS/2 to follow. If Microsoft does not follow
Apple's lead, but instead chooses one font vendor's format, other vendors
will have a hard time translatiApple and Microsoft may also differ in the
markets that they create for add-on fonts. Apple will offer only a small
set (fewer than than 50) of bundled typefaces. All other fonts will be sold
by third parties.
Microsoft is considering selling its own package of the most-popular typefaces,
pre-empting the OS/2 font market. If so, look for a greater range of suppliers
-- and typefaces -- in the Macintosh market. n
Apple and Adobe go their own ways (8/15/1989)
Joel West, "Apple and Adobe go their own ways," MacWEEK,
v3, n30 (August 15, 1989):38.
Although I do my best to interpret the technical side of the Macintosh industry,
there was no time that my interpretation was so in demand as after Apple's
move last month to sell its 16 percent stake in Adobe Systems.
"So what do you think of Apple's move on Adobe?" was the question
I heard from almost everyone I talked to.
Although I can't say I predicted it, the split was the logical and inevitable
consequence of recent developments, not the sudden change in direction that
many of my fellow industry members seemed to think.
From a financial point of view, it was an excellent move for Apple. Even
for a $5 billion-a-year company, an $80 million profit is nothing to sneeze
at.
But the most important question on everyone's mind was the question of what
change would occur in Apple's strategic partnership with Adobe, a partnership
that provided Adobe with instant market credibility and put PostScript on
the map.
Like most strategic business relationships, this one lasted only as long
as the two corporations held goals in common. Companies, like countries,
have only permanent interests, not permanent allies.
The stock sale was the obvious culmination of a divergence that became apparent
with Adobe's failed attempt to railroad Apple into adopting Display PostScript.
Apple declared independence from Adobe last year for its screen graphics
and, with this year's System 7.0 open font architecture, for its screen
and printer fonts.
This confirms what I've been saying for months to anyone who'll listen:
Apple wants to use PostScript as a niche printing solution, but its future
lies with enhancing its own proprietary solutions, such as an extended version
of QuickDraw.
To make this strategy work, Apple must have an alternative available --
whether it is a new version of QuickDraw or a PostScript clone -- before
it can replace the Adobe-licensed ROM it ships with every Laser- Writer
NT and NTX.
And even with a strong QuickDraw offering, the Macintosh is PostScript's
strongest market, and Apple must firmly support the page description language
for the foreseeable future. Being able to proof documents at 300 dpi and
typeset them at 2,500 dpi is crucial to desktop publishing, and it will
be a long time (if ever) before QuickDraw-based typesetters become available.
But using a clone to eliminate the PostScript license fees could reduce
the list price of a LaserWriter by $500 when all of the major laser printer
vendors have introduced or are readying Laser-Writer competitors.
Controlling its own PostScript emulation code will also let Apple design
a printer that supports both PostScript and QuickDraw, an obvious necessity
during Apple's transition to its new technology.
Apple isn't the only one that can benefit from the change. Despite a major
loss of business, the sale seems to be in Adobe's best interests. With such
an obvious divergence of interests, having Apple as a major shareholder
was an untenable position for both companies in the long run.
Adobe had already diversified its sales to reduce its dependence on LaserWriter
sales, with Apple now accounting for less than a third of its revenues.
Now Adobe's challenge is to show that, unlike MicroPro, Digital Research
and Lotus, it is more than a one-product company. The key printer manufacturers
are either signed up or using a clone, and Display PostScript is out of
the running for most platforms. The trick will be to come up with a strategy
that builds from Adobe's strengths but that is not tied to these areas where
its growth is blocked.
The move also has positive implications for all of Apple's investments and
"spin-outs" such as Mirus and Claris.
Apple clearly understands that its role is to encourage its companies to
take the steps necessary to succeed, not to force them to toe the Apple
party line. Assuming it makes prudent investments, Apple wins two ways:
by having the strategic solution that it needs and by the financial gain
from the investment.
Apple needs low-cost Mac as much as we do (8/1/1989)
Joel West, "Apple needs low-cost Mac as much as we do," MacWEEK,
v3, n28 (August 1, 1989):44.
As Apple continues with its strategy to expand market share, a key missing
piece is the low-cost Macintosh.
I would define low-cost as a black-and-white system for less than $1,000,
or a color system with hard disk for less than $2,000. Both need at least
2 Mbytes of RAM to work with System 7.0. It's something developers have
been requesting for years, and Apple is now hinting at plans to reduce product
cost without reducing product capabilities.
Apple's low-end strategy came to mind when I read the report of the Massachusetts
Institute of Technology Commission on Industrial Productivity. Published
by MIT Press and excerpted in the June Scientific American, the report analyzes
why Japanese companies have gained market share in the past 30 years at
the expense of U.S. manufacturers.
In discussing the computer industry, the MIT group notes one mistake that
Apple has taken pains to avoid. Many U.S. companies took a short- term view
by licensing their technology to Japanese competitors. After a few years,
surprise! The U.S. company is being pushed out of the market.
The open standard of the IBM PC AT is easier still to emulate, since IBM
PC clones do not even need a license. Asian clones have almost taken over
the low end, and the leading Japanese companies are trying to establish
market share for higher-end products.
Upon reflection, I must agree with Apple's strategy of not licensing its
technology to competitors. The company's software and firmware (ROM- based
software) are its primary competitive advantage.
But if Apple won't allow others to make a low-cost Macintosh, the company
needs to provide its own, either by reducing the price of an existing CPU
or, more likely, by making a new, less-expensive unit.
Any new Mac must include at least a 68030 if it's not to become obsolete
with System 7.0. The color firmware at the core of Apple's desktop media
push requires at least an '020, while virtual memory needs the '030.
That might describe today's SE/30, which, at $5,000, is hardly "low
cost." How could something similar be offered at a lower price?
The two basic approaches are to reduce the cost of the components and to
make it easier (and cheaper) to assemble those components. Despite a massive
up-front design and tooling cost, a redesigned SE/30 would cost Apple less
to make than today's aging Plus and SE.
In fact, the SE is an example of reduced component costs. Once capital costs
have been amortized, an SE costs Apple less than a Plus because a reduced
chip count provides the same functionality. Motorola is also helping here,
by adding on-chip memory management to the 68030 and floating-point acceleration
to the 68040. Motorola has even produced a chip for General Motors that
incorporates I/O interfaces onto a 68020-like CPU.
However, the low-level Mac design is tied to specific non-Motorola chips,
which must be included or emulated in any future system.
Apple's likely approach is to design its own application-specific integrated
circuits to reduce the chip count of future Macs, much as it designed a
single chip for the Apple IIGS to replace the Apple II chip set.
Recently Apple has focused on designing for ease of manufacturing. As the
MIT study makes clear, a major Japanese advantage has been to design products
suited to a company's manufacturing process, resulting in simultaneous improvements
in quality, cost and the speed of getting products to market.
Low-priced, high-volume products are crucial to this process; with them
a company can realize economies of scale in manufacturing and learn how
to reduce unit costs.
These cost-cutting strategies can be used to reduce the cost of all products,
which is why companies that attempt to keep a high-end niche are vulnerable
to mass producers of low-end, high-quality products. Simpler designs have
fewer assembly errors, and an improved design can more than make up for
higher U.S. wages.
The Macintosh IIcx is Apple's first step in this direction, with a manufacturing
cost much lower than that of either a Plus or many AT clones. Apply the
same techniques to the SE/30 and you would have a low- cost machine that
could be ready late next year.
System 7.0 draws outline of a font solution (6/13/1989)
Joel West, "System 7.0 draws outline of a font solution,"
MacWEEK, v3, n23 (June 13, 1989):46.
Apple's announcements regarding System 7.0 offer an exciting view of the
future Macintosh. One of the most significant breakthroughs is that the
new operating system will provide text-related services.
In previous columns I offered my thoughts on the state of services today
and what new services would look like, and I asked key questions about aspects
of them that didn't appear to be set in stone. For example, six months ago
I asked, "Will huge font libraries of major typefaces become available?"
The news here is good. Apple's new open font standard has garnered an impressive
list of endorsements from ATF, Bitstream, Compugraphic, Linotype, Monotype
and URW.
But converting fonts to a new format usually requires some work by a skilled
font designer who must fine-tune those "hints" for small type
sizes that aren't converted automatically. Such skilled designers are scarcer
than programmers, so the concept of vaporware is as strong in the font world
as it is in software. It remains to be seen how many of these fonts will
actually be ready when 7.0 ships.
In outlining the limitations of then-existing font technology, I also noted
that the scalable fonts used in printer drivers are too slow for screen
display and too poor in quality for 10-point text at screen resolution.
Coincidentally, Apple has displayed 10-point screen fonts to developers
that looked just as good as bit maps and were nearly as fast.
Another important question in any Mac font solution is the role of PostScript
printers and the tens of thousands of PostScript fonts that have been purchased
by Mac users. Last year I speculated about a Mac- based Display PostScript,
but those speculations were premature. In the face of strong Apple resistance,
Display PostScript went instead to Steve Jobs' little cube and never made
it to the Mac.
But a week before Apple's outline font revelations, Adobe announced plans
to offer the portion of PostScript that users most want: its scalable font
technology. When shipped with Adobe fonts this fall, the new Type Manager
software will allow those fonts to be used on the screen, even with non-PostScript
printers. In addition, Adobe is working with Apple to develop a utility
that will translate Apple's font format into fully hinted PostScript fonts.
Adobe has hedged its bets perfectly, working with the new format if it becomes
popular while selling its own format until Apple-format fonts become available.
But this is also one of those rare cases where a big, successful company
with proprietary technology decides to do what's best for the user. If both
pieces of software arrive on time and work properly, Adobe fonts could be
used on the Mac, while Mac fonts could be used with Adobe printer controllers.
In previous columns, I've also talked about kerning, tracking, justification
and line layout. With System 7.0, Apple offers the Line Layout Manager with
built-in support for these typographic necessities that have been separately
implemented by each of the major desktop publishing packages. It also has
a few capabilities I didn't anticipate, such as dynamic ligatures that allow
QuickDraw to draw true cursive characters in real time.
More important, the new manager, coupled with improvements in the Script
Manager, should extend the Mac's lead as an international system architecture.
These changes should make it easier for Macintosh developers to support
key foreign markets.
An example I'm familiar with is the process of line layout, or adjusting
horizontal text placement to allow for differences between screen and printer
fonts. The algorithm my firm developed works with the Roman character set,
which is enough to support the Western Hemisphere and Western Europe.
But such algorithms don't work with other character sets, notably Japanese
and Arabic. Fortunately for us, these markets so far account for less than
10 percent of Mac users.
But these markets should be where Macintosh growth is most rapid. And when
System 7.0 arrives, high-quality line layout for all languages will be built
into the system software for every software package. n
System 7.0: A developer's perspective (5/23/1989)
Joel West, "System 7.0: a developer's perspective," MacWEEK,
v3, n21 (May 23, 1989):30.
At its third annual developers' conference Apple topped all previous efforts
in terms of the sheer volume of information presented.
To some degree, this reflects the dramatically increased complexity of System
7.0, which offers (by quick count) eight new managers as well as revisions
to 16 others.
But among those developers I talked to, Apple won points for its unusual
candor. Warning developers about change as much as year before the new system
ships in a first, as is providing everyone at the conference with a 2-inch-thick
binder containing the detailed interfaces to the system routines.
Does that mean that the "not invented here" attitude and the "I'm
Apple, you're just a developer" arrogance are gone from Cupertino?
Hardly: It is an integral part of the elan or esprit de corps of contemporary
Appledom.
But corporate Apple does seem to have realized that it can benefit from
the front-line experience of the thousands of Mac programmers and engineers
outside the company.
As a consequence, the plan for System 7.0 is responsive to the real needs
of real developers. This is unlike many previous releases, which were designed
in invory-tower isolation or only to meet the needs of Apple's own products.
Beyond the sexiest of the enhancements, such as virtual memory, interapplication
communication and outline fonts, there are dozens of smaller changes that
provide features for which developers have been begging -- or screaming.
Many of these are features that Apple itself has used internally but that
outside developers could not duplicate in a portable, documented way.
One of the most obvious is the floating "windoid" hack used by
HyperCard. Other hacks that will be made kosher include changing Finder
comments, sublaunching, external file systems, automatic background applications
and performance monitoring
These changes also include standard, supported mechanisms to replace techniques
that previously had to be written by each developer, such as tear-off menus,
kerning, ligatures, playing disk-based sounds and real- time multimedia,
not to mention printer and communication drivers.
Overall, I was impressed by Apple's decision to support all machines down
to the Mac Plus, with only virtual memory and the color-related enhancements
unavailable on the 68000-based CPUs. Unfortunately, the 2- Mbyte minimum
means that developers will have to support both 6.0 and 7.0 for some years
to come.
Personally, I was disappointed that Apple isn't moving more aggressively
toward an object-oriented toolbox. But the tools aren't quite ready and
neither is the developer community, so I guess I can take solace in the
official sanction of C++ as Apple's internal language of choice, as evidenced
by its use in rewriting the Finder.
(I was also disappointed to learn at the last minute of Apple's policy of
requiring non-disclosure agreements from developers who are also members
of the press. Forced to choose, I decided it was preferable to leave so
I could write whatever I wanted about System 7.0 in this column and elsewhere.)
Response to developer questions indicated that Apple is working on additional
key features, but their announcement is being withheld until Apple is 100
percent sure it can deliver by the due date. So if the "core"
(mandatory) features run a little late, there will be time to slip in a
few more goodies.
In gauging developer reaction, remember that most developers are specialists,
concerned with one narrow area like fast screen bliting or low-level disk
drivers. Apple's specialists are responding to each of these narrower goals,
guided by a few generalists who are trying to make sure that the pieces
fit together as a coherent whole.
With the plan now established, what remains is, of course, execution. Can
these features be delivered on time? Will they be as good as they are sound?
As the TV correspondents say, the answer remains to be seen.
RenderMan: The PostScript of 3-D imaging (5/9/1989)
Joel West, "RenderMan: the PostScript of 3-D imaging," MacWEEK,
v3, n19 (May 9, 1989):38.
Four years ago, Adobe's PostScript language fueled the explosive growth
fo desktop publishing. Today, Pixar is trying to do the same for 3-D graphics
with its RenderMan interface.
What distinguishes RenderMan from current 3-D technologies is its emphasis
on realism. While today's Macintosh programs display 3-D objects using wire
frames or faceted polyhedra, RenderMan-generated objects appear to be continuous
curves. RenderMan can also simulate realistic surface textures, light sources,
and even camera depth-of-field and motion blur.
A prototype version of RenderMan is already running on the Mac as part of
Pixar's efforts to establish RenderMan as an industry graphics standard.
The role that such a standard could play can be best understood by using
an analogy to PostScript.
RenderMan allows 3-D applications to produce synthetic images that cannot
be distinguished from photographs, much as PostScript allows desktop publishing
applications to produce typeset-quality originals.
In desktop publishing, you use a page-layout program to place text and graphics
in your document, which is an electronic representation of a physical piece
of paper.
Eventually, the program or printer driver translates the document into a
PostScript page description. A PostScript interpreter converts that description
to an image, usually as dots on a piece of paper or typesetting film.
In creating 3-D graphics, you use a modeling application to place light
sources and various shapes. In this case, the document is the electronic
analog of a portion of the 3-D world.
If the application produces RenderMan-compatible output, then a RenderMan
interpreter can convert that output to light intensities in a video signal
or a 35mm slide.
Like PostScript, RenderMan is device-independent, allowing the application
to produce output without knowing how it will be displayed. It is also resolution-independent,
so users can resize and reshape images at any resolution without introducing
defects.
With some minor extensions, the RenderMan file format could be used to support
a 3-D clip art library, much as Encapsulated PostScript (EPS) defines a
standard way to exchange two-dimensional PostScript art.
As Adobe is with PostScript, Pixar is more interested in getting other developers
to support RenderMan output than to interpret it. Pixar intends to sell
its own RenderMan interpreter, called Photorealistic RenderMan, based on
its previous-generation REYES (render everything you've ever seen) software.
Unlike Adobe, Pixar has clearly spelled out from the outset what rival interpreters
must do to be called "RenderMan- compatible."
Autodesk has provided the first key Macintosh endorsement by announcing
that its Autoshade rendering option will be based on RenderMan. Other developers
are studying the format's specs, with further announcements likely.
But although Pixar sells Photorealistic RenderMan interpreters for Sun workstations
and 386-based PCs, it does not yet offer developers an interpreter to run
on an ordinary Macintosh II.
Levco has ported Photorealistic RenderMan to the Macintosh but requires
a $4,000 configuration of its Transputer NuBus board to do the computations.
This provides better performance than a Macintosh-based solution but raises
the entry-level cost to developers and users.
How does RenderMan affect Apple's plans for 3-D on the Mac?
Apple has always based its Macintosh sales on proprietary software technology,
and its Advanced Technology gruop has been furiously hiring Silicon Valley's
3-D experts to build 3-D technology for the Macintosh. The first evidence
of this effort was the Pencil Test video produced for last summer's Siggraph
conference.
Thus, it's not surprising that Apple was not among the computer vendors
endorsing RenderMan at its unveiling last year.
But if RenderMan should establish itself as an industry standard that can't
be ignored, then Apple would be forced to redirect its efforts to work within,
or along-side, that standard.
RenderMan certainly has a head start over any proprietary Apple solution:
its first Academy Award. This year's Oscar-winning animated short subject,
"Tin Toy," was produced on Pixar equipment using Photo- realistic
RenderMan.
Peripherals will be on center stage at developers'
conference,(4/25/1989)
Joel West, "Peripherals will be on center stage at developers'
conference," MacWEEK, v3, n17 (April 25, 1989):88.
In less than two weeks Apple will convene its third annual developers' conference,
proferring its strategic vision for the future and answering nitty-gritty
questions about the present.
One of the hot topics on the agenda will be mass storage peripherals. Those
who support such products say Apple has two opportunities for the future:
improving performance (particularly a multitasking operating system) and
providing support for foreign file systems.
The cornerstones of Macintosh peripheral performance are the SCSI Manager
and related hardware. While SCSI has fueled the availability of a wide range
of peripherals, its current performance is inadequate for larger disk drives
and a true multitasking operating system.
At the core is the restriction that all I/O transfers must be made by the
processor. In its initial SCSI implementation on the Mac Plus, Apple tried
to speed I/O by providing "blind" transfers. however, that version
was extremely timing dependent, so that Mac Plus blind transfers don't work
reliably with all disk drives.
The SE, II and later machines include hardware changes that assure transfers
are not made unless the data is ready. This also pushes the maximum SCSI
throughput from 263 Kbytes per second on a Plus to 656 Kbps on an SE. The
Mac II is twice again as fast.
But unlike most high-performance computers, the Mac doesn't provide direct
memory access, which would allow special chips to perform data transfers
without tying up the CPU. Third-party SCSI boards show that this can double
the effective throughout of a Mac II to 3 Mbps.
While such performance is wasted on most of today's hard disks, it is sorely
needed for 300-Mbyte and larger hard disks and for next-generation drives.
Also lurking on the horizon is the issue of multiasking performance. On
a single-tasking operating system, tying up the CPU isn't all that important.
But on a multitasking OS, the SCSI transmission wastes horsepower that could
be used by other tasks.
More significantly the Macintosh spends most of its time waiting for the
disk head to position, not transferring data.
First on every developers wish list is modifications to the SCSI Manager
and the hierarchical file system (HFS) that will allow asynchronous operation.
This would enable disk drivers to sort disk accesses by position on the
disk, as has been done for two decades on larger computers. Such sorting
would speed up some tasks twofold to fivefold. For example, you'd see an
increase when the Mac rebuilds the desktop after a crash, task that can
now take 10 minutes on an optical disk.
Apple's other opportunity lies in reading and writing non-Mac disks. As
with most computers, the Mac has its own standard disk format. Unlike many
other such formats, HFS can also be extended to support other disk formats,
thus allowing the Finder and applications to read those disks as if they
were a Mac disk.
On such extension is the Apple Filing Protocol, in which the AppleTalk network
is used to access a remote disk, even one that is hooked to a DOS or VAX/VMS
computer.
Apple already supports read-only direct connection of one foreign file system
on its CD-ROM player, with a Startup docuent (INIT) that allows any Mac
program to read a High Sierra CD-ROM using HFS calls.
Similarly, a read/write HFS foreign file system for MS-DOS would let users
take full advantage of the 1.4-Mbyte SuperDrive included in the new Macs.
Today, users can only read and write DOS disks using the separate Apple
File Exchange.
Apple already has a prototype of the necessary software but hasn't released
it yet. Why not?
One possible argument is that Apple's distribution of the software would
threaten large installation sales, even though third parties already provide
this capability. How is it a threat? When one or two users need a Macintosh
(such as for desktop publishing), Apple usually sells machines to the entire
work group. If those few machines could easily read and write DOS floppies,
the argument goes, perhaps MIS would buy all the other users cheaper PC
clones.
I, for one, don't buy it. For example, the Mac's training advantages are
as strong as ever.
But until Apple sees this as more of a user advantage than a sales disadvantage,
this is one piece of extremely useful technology that will remain locked
up in the lab.
More from Apple equals less from third parties?
(4/11/1989)
Joel West, "More from Apple equals less from third parties?"
MacWEEK, v3, n15 (April 11, 1989):40.
When Jean-Louis Gassee moved to California, Apple acquired a controversial
product champion to replace Steve Jobs.
On Gassees watch, Apple has gone from one machine, the Mac Plus, to a broad
range of workstations comprising six different models. That transition is
extremely important to the company's financial health and stability.
But the climate for outside development of Mac-related hardware is now less
favorable. If Apple took two steps forward in 1987, it took one step back
in 1989. The opportunities provided by the "Open Mac" have been
diluted by a combination of competition and confusion.
The introduction of the Mac SE and the II in 1987 created a large market
for hardware and-on products while providing Apple with a stronger machine
for business and a high-end flagship. The one weakness was a huge gap between
the SE and II in capabilities and price.
The addition of the IIx, SE/30 and IIex fills in that gap and effectively
defines two product lines: compact and modular. Now there is a real question
of which of which machine to buy. A year ago, if you wanted color or 68020-based
machine, the automatic choice was a Mac II.
Even with that expanded product line, the cost of key features hasn't changed
much.
Suppose, for example, that you want to buy an entry-level color system.
In 1987, the list price of a Mac II with 1 Mbyte of RAM and a 40- Mbyte
hard disk was $6,996. A similarly configured IIex system costs -- surprise
-- $6,996.
What's interesting here is that fully configured systems cost about what
they did two years ago, but models without hard disks are more expensive.
This puts pressure on third-party disk manufacturers as Apple seeks to expand
beyond its 30 percent share of the drive market.
Apple is also increasing pressure on third-party accelerators and video
systems.
SE accelerator makers are hit by the new SE/30, which costs only $1,000
more to buy than an SE. Retrofitting an SE with an SE/30 motherboard costs
about $1,700, and for that you get Color QuickDraw to boot.
At the same time, Apple's new black-and-white monitors for its Macintosh
II series could force most products out of that market.
THis is the same approach used by IBM and Digital Equipment Corp.: Wait
for third parties to create a market, and if it's big enough, try to take
it over.
To be fair, Apple has always said it will compete in hardware. Those manufacturers
that want to be around for the long haul must stay two steps ahead of Apple.
"There's a big foot that comes along every so often in your market,"
said one entrepreneur. When you see the Apple logo on the bottom of a foot
about to stomp on you, he said, it's time to ready other products.
What other products are these companies likely to offer? Not NuBus color
video. The market there is already crowded and is likely to see an Apple
introduction next year.
Apple is encouraging boards for the SE/30 and seems to be steering clear
of the SE/30 add-on market, as it did with the SE. Apple also said there's
another CPU coming that will be compatible with SE/30 cards.
Adapting the hardware of an SE/30 board from a NuBus board requires only
about 25 percent more work. In addition, the more-costly firmware moves
over for almost nothing.
Even with these attractions, confusion remains. It the "'030 Processor
Direct Slot" important enough to support a wide range of products?
Will dealers and value-added resellers stock three types of cards, one each
for the SE, the SE/30 and the NuBus-based machines.
How will the Mac SE/30 be used? Mac SE expansion was mostly video cards
and accelerators, while the Mac II inspired a more eclectic range of offerings.
Will engineers buy an SE/30 or IIex for data acquisition?
Overall, which of the two machines will be more successful? When I asked
John Sculley at the IIex press conference, he told me that Apple had no
internal forecast on that question.
I suppose Aple could be using big-Blue's approach for introducing a wide
variety of new products: Throw everything against the wall, then wait to
see what sticks.
Or perhaps, as Jean-Louis Gassee might have suggested, Apple does have a
forecast -- but it's from the same squad that predicted DRAM demand.
At the trailing edge of technology (3/28/1989)
Joel West, "At the trailing edge of technology," MacWEEK,
v3, n13 (March 28, 1989):62.
In product announcements, the weekly headlines or even a bull session over
pizza, the tendency is to focus on the latest and greatest products.
But the reality of a maturing Macintosh market is that many users do not
have the latest and greatest products and are, in fact, quite happy to remain
on the trailing edge of Macintosh technology. They're sticking to their
old reliable computer, system software, disk drive, printer and applications.
The phenomenon is not just found in Dubuque or Klamath Falls but in big
cities, big corporations and big universities.
Ironically, many of these trailing-edge users were early and vocal supporters
of the Macintosh. Because they were early adopters they still have the older,
less sophisticated equipment while more recent converts make use of state-of-the-art
hardware and software.
I know of several cases in which an original Macintosh 128 became a 512
and then a Plus. Some have accelerators but many Macs remain underpowered
and useless. That is, if you believe the "leading edge" adherents.
At larger sites, the upgrade cost is multiplied by a factor of 10 or 100.
This is particularly noticeable in universities that populated computing
laboratories with Macintoshes early on.
Cost is not an issue for system software, since updates are free if you
can find a dealer or users group that will make you a copy. Still, you'll
find many older systems and Finders on these machines. For example, I'm
writing this article under System 4.2 and Finder 6.0. MultiFinder is an
option only for those with more than 1 Mbyte, still a minority among installed
machines.
Although many of these users might be suffering from Mac envy, there are
also logical reasons for life at the trailing edge. At larger corporations,
trailing-edge technology might be an integral part of the company's computing
strategy.
I recently spoke with an engineer at a large aerospace company who asked
me about using Switcher, a program I always found clumsy and haven't used
in years. The company's MIS department had only recently distributed MultiFinder
-- 18 months after it was first released.
For the larger sites, the initial bias is to delay upgrading unless the
benefits are overwhelming. There is a training cost associated with new
software. And, as the managers of mainframe computers learned long ago,
the first users of any new version of an operating system are guinea pigs.
Some system managers would test a new mainframe operating system on just
one machine, but this usually isn't practical in a networked Macintosh shop.
For one thing, all systems sharing a LaserWriter printer must use the same
version of LaserPrep.
In some cases, the impetus to upgrade comes only when a user or site selects
an application that requires it.
Apple has traditionally encouraged developers to target the new machines,
claiming that those who use older technology are least likely to buy new
software. Their requirements are static and, besides, if you can't spring
for a more powerful machine, how can you afford an $895 software package?
On the other hand, if your problem is static and your existing solution
works, why change? Although I have the newer versions, I prefer MORE 1.1
to MORE II, and SuperPaint 1.1 to SuperPaint 2.0. The only time I feel compelled
to upgrade is when the previous version is buggy and poorly done.
Apple has made a significant effort to keep system software up to date for
one of its most popular systems, the Mac Plus. This should remain a viable
(although trailing edge) platform for several years.
So if you don't have the latest and greatest, don't despair. You're far
from alone.
Engineering: More than just CAD (3/14/1989)
Joel West, "Engineering: more than just CAD," (column) MacWEEK,
v3, n11 (March 14, 1989):38.
Within the engineering market, Apple's interest is in themost wide- ranging
solutions, those that sell thousands and tens of thousands of computers.
For example, if you classify a Mac II as a workstation, then Apple sold
more workstations in its first six months than Sun Microsystems sold in
its first six years.
Thus, it makes sense that the Mac engineering focus has been on the most
horizontal of the engineering applications, CAD.
The Mac's strong graphics orientation makes it a more natural CAD platform
than most computers. MacDraw and other drawing programs showed the Mac's
potential even on the early underpowered machines.
Once it had the Macintosh II, Apple's first engineering targets were CAD
and CAM. Incorporating the related fields of mechanical engineering, civil
engineering and architecture, CAD/CAM represents one of the largest and
best-understood engineering markets.
Today, many CAD packages are available for the Macintosh. Several are popular
enough to be found at the local computer store. But such general-purpose
CAD solutions cannot solve every drafting problem. Each specialized area
requires different features, either as part of a horizontal CAD package,
as a separate add-on or as part of a more narrowly focused "vertical"
solution.
For example, an architect needs a 3-D CAD package that will generate drawings
using standard construction symbols. Because architecture is the primary
market for many 3-D CAD packages, such support is often an integral part
of these packages.
On the other hand, surveying has one fundamental difference from general-purpose
CAD: Real-world surveying measurements have errors that must be averaged
across the drawing to produce polygons that meet at each corner.
There's more to design than just drawing, and Apple has attracted an impressive
array of structural analysis tools, such as skeletal and finite-element
analysis.
Other related engineering solutions available include conversion of CAD
data to standard CAD formats like DXF, IGES or CGM, with nine-track tape
and Ethernet for transporting that data. Other key peripherals include 3-D
input devices and E-size plotters for bluprints.
With Apple's marketing efforts for structural engineering in place, its
latest engineering interest has shifted to electronic design automation
(EDA).
Today, the Macintosh is weak in two key areas of basic electronic design:
schematic capture and printed circuit board layout.
The fomer is a form of analysis that allows the designer to build the circuit
schematic, while the latter is a specialized form of CAD that places individual
components on the tiny PC board real estate.
A complete board-level design package can export the schematic database
to the layout program, auto-route connections between components, validate
the layout against the schematic and prevent obvious mistakes, such as wiring
one output lead to another. Several such packages are available on DOS machines,
but none of the Mac solutions yet fill the bill.
Other important EDA tools include cross-compilers for microprocessors and
logic equation compilers for designing PALs.
Apple itself is aware of the limitations of Mac-based solutions, and it
recently has purchased EDA systems for its own engineers from dedicated
CAD systems vendor Mentor Graphics. Mentor's systems comprise its proprietary
software with high-end workstations from Apollo.
This illustrates two growth areas for Mac engineering solutions.
First, while Apple may aggressively evangelize the largest CAD markets,
value-added resellers like Mentor control most of the medium-size niches
in the engineering market. Platform choices by these resellers ultimately
determine which Mac-based solutions are offered to engineers.
Second, Apple's current computers cannot support the most computationally
intensive design applications. Until faster products are available, many
engineering markets will remain closed to the Macintosh.
Desktop engineering playing from strengths (2/28/1989)
Joel West, "Desktop engineering playing from strengths," MacWEEK,
v3, n9 (Feb 28, 1989):62.
In its attempts to crack engineering and related technical markets, Apple
has a number of winning cards in its hand.
Apple's recent marketing efforts have stressed that, in effect, engineers
are people, too, people who spend most of their day doing work other than
hard-core engineering.
Engineers write memos and reports, prepare presentations and fill out expense
reports. In a small company, they might share a laser printer with accounting.
In a large company, they must download information from the corporation
mainframe and track project results.
In the past five years, the Mac has proved itself more than adequate for
such purposes. As with DOS microcomputers, the hardware and software are
cheap compared with more expensive workstations, minis and mainframes. Micros
are readily available through a wide variety of convenient retailers, thus
making it easy for many companies to justify buying such computers without
a specific engineering usage.
In some ways, these comparative advantages also reflect weaknesses. Back
in the days of the underpowered 128K Mac, about all that engineers could
use their Macs for was to write cute memos in MacWrite. Even today, the
Mac and Mac-based solutions are only cracking a small part of the computer-aided
engineering market.
At the same time, the winning attributes of the Macintosh in business markets
provide the cornerstone of Apple's efforts to expand sales in technical
markets.
The common thread for computer-aided engineering has been complex quantitative
analysis: "number crunching." This evolved from slide rules and
stacks of scribled notes into the reams of computer-generated tabular data
of the 1960s. Next came the ability to display graphical representations
of such complex data on the screen. The Mac's role in this arena is already
well-established.
Of course the Macintosh's strongest suit has been its ease of use. That
is now becoming a key requirement for engineering computers, in hopes of
reducing the relearning time required by engineers after long gaps between
each new design or analysis effort.
The Mac's graphics and ease of use fueled the zeal of that species of early
adopters known to friend and foe as "Mac fanatics." The fanatics
have played key roles in selling the Macintosh in the large organizations
included in the MacWEEK, 200, such as the major aerospace companies
represented there.
But Apple's most important engineering move was introducing the Macintosh
II almost two years ago. At last, the Mac was competitive with high-end
DOS machines and low-end workstations, correcting a performance disadvantage
that was particulary acute for floating-point numerical calculations.
Still, the current Macintoshes represent little threat to established engineering
and graphics workstations for real-time data acquisition and simulation.
And due to the Mac II's relatively late entry, both DOS and workstation
platforms have a large head start in attracting engineering solutions. Apple
is working hard to address that lead.
Before the Mac II, Apple by-passed the annual A/E/C engineering trade show.
It has been a major presence there ever since.
To complement such marketing efforts, Apple recently made changes to improve
the distribution and support of Macintoshes to engineering companies.
As previously adopted for its A/UX efforts, Apple has pre-qualified a fraction
of its reseller network to target engineering sales. The Apple- provided
specialized training won't turn salesmen into rocket scientists, but it
will better enable them to identify off-the-shelf solutions appropriate
for each customer's needs.
And as part of the recent reorganization of the Apple Developer's Group,
Apple now offers clearly defined support plans for two other conduits to
engineering buyers. Value-added resellers have been integrated into the
Apple Partners category, on a par with certified developers. And computer
consultants -- those who provide specialized advice without actually selling
hardware -- have their first official support from Apple as Apple associates.
Still, as with other Apple efforts to reach more customers, the company
needs an ever-wider range of specialized solutions to win in the engineering
market. I will discuss that need in my next column.
Will every user become a programmer? (2/14/1989)
Joel West, "Will every user become a programmer?" MacWEEK,
v3, n7 (Feb 14, 1989):34.
DO YOU WRITE SOFTWARE for the Macintosh? Even if your employer isn't Microsoft
or Claris, Apple may be looking for you.
Apple's newly restructured Apple Developer's Group is now reaching out to
Macintosh users beyond its original target of third-party software developers.
This comes as more programming power is being placed in the hands of users.
Apple's original Certified Developer program was responsible for the wide
variety of software and hardware available for the Mac in early years and
has been widely copied by Apple's competitors.
Such a successful track record is not something to be changed lightly. Why
then did Apple announce a revamped program at last month's Macworld Expo?
Previously Apple's attention was focused on developing retail software that
sells thousands of copies. Such software serves the needs of most Mac owners,
particularly the individuals and small businesses that were the Mac's early
users.
But recent years have seen the Mac reach large corporate and government
organizations. At many of these sites, custom applications are developed
by in-house programmers who previously had little help from Apple.
In addition, user programming traditionally has been important in the scientific
and engineering markets, which have been targeted by Apple. Although an
engineer may use a pre-programmed CAD package the way an accountant uses
a spreadsheet, a technical researcher is more likely than most users to
write new, specialized code to solve particular problems.
How has Apple addressed these changes? The new Partners and Associates program
aims at two categories. The "partners" designation subsumes commercial
software developers and publishers, hardware manufacturers and value-added
resellers.
What's new is the extension of benefits to "associates," which
includes anyone who develops software for his or her own organization, particularly
those in education and the Fortune 500. The program also picks up those
who previously fell between the cracks, including consultants, book authors
and industry analysts.
Both categories get technical documentation and other mailings. The Partners
get hardware price breaks and AppleLink tech support.
The pricing is designed to favor large organizations and those who are seriously
committed to the Mac. Partners pay $600 per year vs. $350 a year paid by
Associates -- a far cry from the early, disorganized years when a $100 software
supplement subscriber received more than that back in floppy disks.
At the same time the definition of programming has grown from traditional
application programming to include application customization. The idea,
pioneered by products such as dBase II and Lotus 1-2-3, is to provide a
way for users to build upon the built-in functionality of an application
rather than build that functionality from scratch.
A database or spreadsheet would now be considered inadequate without programmable
macros. Such capabilities are extending into other areas, such as communications.
This may range from simple log-on macros to an integrated Mac-like front
end to a mainframe database, as provided by Apple's scriptable MacWorkstation.
Apple's most important break-through here has been HyperCard, an intelligent
browser in which the HyperTalk programming language was almost an afterthought.
It would be wrong to say that any user can become a HyperCard programmer,
but the package is more accessible than any previous programming environment.
HyperCard has created a whole new industry of third-party solutions, particularly
in vertical markets.
Nonprocedural programming has also become important. Here, users describe
relationships instead of algorithms. Nonprocedural programming is manifest
in many products, from spreadsheet formulas, relational databases, style
templates and communication profiles. Such description are also the basis
of some forms of artificial intelligence, including expert systems.
An increase in user customization does not mean a return to a decade ago
when hobbyist computer owners wrote all their own programs. Instead, more
elaborate customization through computer-aided software engineering promises
to replace today's laborious process of application development.
Apple's tool development is important cog (1/31/1989)
Joel West, "Apple's tool development is important cog," MacWEEK,
v3, n5 (Jan 31, 1989):32.
APPLE'S EVANGELISTIC EFFORTS on behalf of tools used for software development,
such as programming languages, is unique. It both encourages outside products
and preempts portions of the marketplace with its own products.
The right tools are an important lure for writing new application software,
as well as a factor in making that software available sooner to users. What's
more, Mac software development has expanded beyond just the commercial developer
to include more end users. (I'll discuss this in more detail in my next
column.)
As with application software, when publishers of third-party tools make
money selling their solutions, Apple makes money selling hardware, and users
have a wide range of off-the-shelf solutions to choose from.
Third-party tool developers get encouragement from Apple, but their products
can also be wiped out if Apple competes. As the joke about IBM goes, if
you get in bed with an elephant, don't expect to sleep very well because
the elephant may roll over and crush you.
Why does Apple develop such tools? Originally, all software was strategically
important to the Mac, but now Apple's focus is limited to "system software."
And that includes tools.
Most computer manufacturers have three goals in their tools development:
* To facilitate development of operating system software.
* To let others write software.
* As a separate profit source.
Apple's tools reflect its open systems interconnect requirements, including
coding new ROMs, diagnostic programs, System, Finder and drivers. The original
system software was developed in assembler and Pascal, but the recent emphasis
has been on C and C++, which is why all are found in MPW. And many of the
revision control features of MPW 3.0 stem from Apple's difficulties in testing
ROM and System updates.
Apple's tools match its original focus on developers of retail business
applications, such as word processing and spreadsheets. Again, the availability
of C, Pascal and assembler meets such needs.
However, the new thrust into scientific and engineering markets assures
a crucial role for other languages. These less-popular niche languages have
so far been handled exclusively by third parties.
Apple recently bought one such company, Coral Software, which offers Common
Lisp. But Apple's goal was to start the nucleus of an AI research lab in
the midst of the Cambridge/Route 128 talent pool, not to assume responsibility
for Coral's existing products.
Besides Lisp, other important research and engineering languages include
fortran, Ada and Smalltalk, and in all cases, Apple depends on outside suppliers.
Fortran development has been entirely driven by third parties, as Apple's
young staff has little experience with this aging but still popular language.
Apple's recent thrust into the federal market has lured Ada vendors Altsys,
Meridian and Telesoft to develop this key product for government bids.
Meanwhile, Apple killed its own nearly complete MacSmalltalk 2.0 in favor
of Digitalk's Smalltalk/V. The fact that the PC Smalltalk/V is the only
commercially successful object-oriented product was an important factor
in that decision.
The small size of the tools market means there will be no smash hits comparable
to an Excel or PageMaker. To date, there are only two clear success stories.
One is TMON from ICOM Simulations, a debugger that is more popular than
Apple's MacsBug.
The second is LightspeedC, the most popular Mac compiler. Other C suppliers
are either gone, or lag far behind. In contrast, Pascall dollars are split
between MPW, Lightspeed, Turbo and TML.
Apple has helped third parties by licensing MPW, which is included with
products such as Language Systems' Fortran, Meridian's Ada, SemperSoft's
Modula-2 and TML's Pascal. That Apple did not charge hefty license fees
to recover its multimillion-dollar MPW development cost suggests the strategic
importance of such outside products.
But economic realities ensure that Apple will continue to be active in providing
tools. Such tools are the foundation of Apple's strategy of using software
to distinguish itself from its hardware competitors.
As one insider confided, "If we left it to other companies, most would
go bankrupt, and there wouldn't be anything being done."
Five great moments in Macintosh history (1/17/1989)
Joel West, "Five great moments in Macintosh history," MacWEEK,
v3, n3 (Jan 17, 1989):102.
THIS IS the time when many of us pundits reminisce about everything that
has happened since the Macintosh was introduced five years ago.
Coincidentally, it's also the time of year for the annual NFL end-of- season
frenzy.
The two audiences are not as unrelated as they might seem. At last year's
Macworld Expo in San Francisco, in fact, attendance was markedly down during
the 49ers playoff game.
For the football fan, one of the most popular TV features in the past few
years has been the flashback to a great moment in football history, replaying
a great touchdown run or interception return.
I decided I'd like to try something similar in honor of the Mac's fifth
birthday, looking back at how Macintosh technology -- and the benefits it
offers -- has evolved in those five years.
The Mac has come a long way since that first "insanely great"
yet underpowered machine of 1984.
The original Macintosh was conceived as a home appliance, like a toaster,
that would be used by individuals, not companies. As a side bet, Apple mounted
an aggressive effort to convince each and every college student to own a
Mac. Both were logical approaches for a company whose biggest markets were
home computers and K-12 education.
Since then, the Mac has become a serious business computer, establishing
its beachhead in corporate America and changing the rules as it goes along.
I've arbitrarily chosen five steps along the path from there to here. In
chronological order, they are:
1. LaserWriter (1985). It saved the Macintosh, and possibly even
the company, at its low ebb. It made PC owners insanely jealous and launched
the Mac as a publishing platform, while inspiring the crucial second wave
of Macintosh software.
And for a long, long time, it was the only reason to use AppleTalk -- reason
enough to make AppleTalk the single most successful networking architecture
as measured by the number of installed nodes. It also gave AppleTalk the
field test that any product needs to evolve and mature.
2. Excel (1985). Some might argue for PageMaker, which allowed people
to do things they had never done before. But for the operations side of
corporate America, it was more important to do something it understood (bean
counting) better than was previously possible.
The hard-core techies knew that a 68000 was more than a match for an 80286,
but it took a head-to-head comparison to prove it.
3. Macintosh Plus (1986). Sure, it had more memory than an IBM PC
AT and a decent-capacity floppy drive. And providing the option of a fast
hard disk removed the last remaining obstacle to corporate acceptance.
More important, the Plus was the first hardware that was legally expandable
by third parties, a step toward an open hardware architecture. You could
now legally add third-party memory or a third-party disk drive to any machine.
The fact that SuperMac, rather than Apple, dominated the early SCSI drive
market cemented the role of third-party hardware suppliers in the Macintosh
market, guaranteeing a wider range of alternatives for the user.
4. Macintosh II (1987). The first of a second generation of CPUs,
the Mac II introduced color and gray-scale capability, not to mention real
horsepower.
The NuBus gave Apple a bus standard for the future. Since then, the range
of third-party expansion options has proved that the Macintosh II is both
a desktop microcomputer and a low-end workstation.
5. HyperCard (1987). It spawned hundreds of new companies, gave Apple
a way to use CD-ROMs, made Danny Goodman a celebrity and forever changed
the defintion of "system software." Meanwhile, its real impact
has only begun to be felt.
My only question is, what can Apple Fellow Bill Atkinson possibly do for
an encore?
In other areas, the evolution has been spread out among several products
or revisions, rather than coming at one fell swoop.
For example, MultiFinder 1.0 was but one step, albeit crucial one, between
Switcher and a true multi-tasking OS that has yet to appear.
And the Kanji Mac was a prototype for the truly international, Script Manager-based
Macintosh of today.
As for the next five years, I'm not ready to prognosticate just yet. I'd
probably have better luck predicting who will win the 1994 Super Bowl.
The '030's advantages for business users (1/3/1989)
Joel West, "The '030's advantages for business users," MacWEEK,
v3, n1 (Jan 3, 1989):34.
APPLE WILL be introducing a series of 68030-based systems later this year,
according to a recent speech by CEO John Sculley. Is this something the
average business user should care about? For the following reasons, the
answer is yes.
The current Macintosh Plus and SE contain the same Motorola 68000 CPU as
the original 128K Mac, although its performance has been enhanced slightly
because of changes in the video circuitry.
What user advantages does the 68030 offer over the 68000? To answer that,
we need to look at each generation succeeding the '030.
The first successor, the 68010, corrected small problems in the original
Motorola design and allowed it to support a multitasking operating system.
The chip formed the heart of the first generation of engineering workstations.
Next came the 68020, which offered about four times the horsepower of a
comparable 68000. It is considered the minimum necessary configuration to
support Color Quick-Draw, the cornerstone of Apple's color and gray- scale
graphics architecture.
Lastly, the 68030 integrates memory-management functions onto the CPU, allowing
an operating system to protect one application from "bombing"
the entire computer.
Why choose the newer '030 (now found only in the Macintosh IIx) over the
'020 of the Macintosh II? Not performance, which is only about 20 percent
faster than the '020, and that primarily because of an added cache. No,
the main advantage of the newer '030 is the on-chip memory- management unit,
or MMU. An MMU enables the CPU to assign different parts of RAM to separate
applications, preventing one program from tampering with the memory used
by another.
Motorola, for one, would like Apple to switch to its latest chip as of a
head-to-head fight with Intel for prestige and market share.
The Apple account is important to Motorola's semiconductor division, both
symbolically and financially. One brokerage estimated that one out of every
seven 68020s shipped last year resided in a Mac II, making Apple Motorola's
flagship customer.
By adopting the 68030 over the 68020, Apple indicates that it believes there
is an advantage to increasing the installed base of systems with an MMU.
A comparable situation occured a few years back in the MS- DOS world. Microsoft
knew that its OS/2 would require the memory protection first available with
the Intel 80286. Experience has also shown that Windows and the OS/2 Presentation
Manager require the performance of an 80386- or fast 80286-based system.
Before introducing its new operating system, it was important for Microsoft
to encourage sales of '286- and '386-based machines. That strategy has suceeded,
with most of the MS-DOS machines sold in the past two years containing CPUs
ready for OS/2.
By the same token, a future version of MultiFinder that supports memory
protection and true multitasking requires the Mac to have an MMU. This can
be either the built-in variety of the 68030 or the external 68851 available
for the 68020. If most of the systems sold in 1989 and 1990 provide such
an option, the Macintosh world will be ready, much in the way of the IBM
PC and compatibles world, for an improved operaing system, which Sculley
said is due in 1991.
In addition to memory protection, an MMU provides support for a virtual
memory operating system. This allows an applciation to directly use more
memory than is physically available in RAM. You could use a 2- Mbyte machine
to manipulate high-resolution images, large page-layout documents or complex
CAD drawings.
A new generation of hardware is also due for math calculations. We probably
won't see the 68881 floatingpoint coprocessor on any machine after the Macintosh
II. Instead, the 68030 is normally paired with the faster 68882 math coprocessor,
as is found in the Mac IIx.
For a comparable price, the '882 provides a theoretical 50 percent speed
improvement over the '881. Motorola has discouraged new applications of
the '881 and plans faster chip cycles only for the '882. All of this points
to why users should care about any new 68030-based platforms.
If you like, you can try a 68882 today, as it is a pin-for-pin successor
to the 68881 and can be used to replace that chip on any Macintosh II.
The external math coprocessor should disappear entirely with the built-in
floating point of the 68040, which Motorola plans to introduce later this
year. But it probably will be 1991 before the Macintosh crowd reaches fortysomething.