Ten years ago to this very week, Steve Jobs killed Apple. Or he began to take apart many of the projects and organization that many inside and outside the company thought of as Apples value to the computing industry.
This event was the announcement of the companys infamous spring 1997 reorganization, which continued step-by-step throughout the spring. For the companys long-suffering developers and ISVs, push came to shove at the annual WWDC (Worldwide Developers Conference), where Steve Jobs revealed his plans for Apples future direction.
Today, with the iPod-influenced haze over of recent Apple history as well as the success of Apples retail strategy and the Intel-Macintosh transition, we forget that some of that “future” talked up in 1997 never happened.
WWDC and Apples relations with developers, past and present, are on my mind. And no wonder.
Apple media representatives this week were ringing the phone making sure that reporters are entering the June dates for the 2007 WWDC in our calendars.
The hubbub on the Mac enthusiast sites circles around the actual release date for Leopard, Mac OS X 10.5, and the expected concurrent release of iWork 07, an expanded edition of Apples productivity suite. Theres a lot of talk about an as-yet-unannounced special Apple event coming on Feb. 20.
Whatever the details, Apple had better have shipped Leopard before the next WWDC. Its already talked up the updates future and new features at two past conferences; it cant have deja vu all over again.
Apples WWDC story to developers must be how they can write (and sell) cool new versions of programs that will leverage the new features. Of course, developers will only buy into that message with a whole heart if theyve seen the lines queuing up to get Leopard. (And be sure, unlike the fizzled turnout for Windows Vista promotions, Mac users will be out in force to install Leopard, just as they did for the current Tiger version.)
The company also has to shape up its tools. In the fall, Apple said it was adding XRay, a dynamic application performance framework based on the OpenSolaris DTrace.
However, for a long while now, developers have complained that the introduction of new features has outpaced the fixing of bugs, especially those from the new features.
Recognition of this situation can be seen in a Tuesday, Feb. 6, posting in the Surfin Safari blog by programmer Maciej Stachowiak. This blog covers WebKit, the framework used by the Safari browser, the Macs Dashboard widget applets and other Mac OS X applications.
“After many discussions with interested parties and members of the WebKit community, weve decided the time has come to get serious about stabilizing the code. Weve had about two years of development, which has included many awesome compatibility fixes, performance improvements, standards compliance enhancements, and new features. Now is the time to get the source tree in solid shape,” Stachowiak wrote.
Developers say they cant wait.
“This is a good thing. WebKit has been kind of running wild with new projects going on—removing dependency on the Mac OS, adding SVG (scalable vector graphics) support and more. Thats all well and good, but there are tons of bugs that need to be fixed,” said Dan Wood, co-owner of Karelia Software, of Alameda, Calif. The company develops Sandvox, a simple, real-time Website editing tool.
“Clearly Apple is trying to whip WebKit into shape for a Leopard release,” he added.
Of course, the whole world will be watching this year, with Windows Vista finally on the market. Apple has had the platform comparison scene easy for the past few years thanks to Microsofts nightmare development process.
While Apple has revealed some of Leopards features in its “sneak peak,” even the most loyal developers are waiting on the entire package. When Leopard was announced, Jobs said Apple was keeping a number of features secret for its release.
Some of these “hidden” features are already revealed, such as resolution independence for displays, which decouples the resolution of OS elements, such as windows and menus from the physical pixel density of computer screens, and integration of ZFS (Zettabye File System), which is included in Solaris 10.
These forthcoming Leopard capabilities are not hyped on Apples site. Still, developers expect them and theyre mostly working in the Leopard builds.
Leopards Flash Cache Secret
But for Apple, a software feature can also be a hardware feature. Heres my own very circumstantial conjecture to support another rumored Leopard feature.
Back in the summer, I wrote about Intels progress on Robson, the companys forthcoming flash cache architecture. Supporting Windows Vistas ReadyDrive and ReadyBoost acceleration technologies, Robson offers an alternative to the use of “hybrid” flash-enabled hard disks and external flash thumb drives. The architecture comprises a flash cache of up to 4GB, a controller ASIC on a PCI Express minicard and driver software on the host computer.
There are several advantages that Robson has over the use of thumb drives and hard disk caches for this system acceleration. First, because its placed inside the computer, its considered “safe” and the system trusts its data integrity between sleep states or even shut downs. Vista rebuilds the cache on a thumb drive in between sleeps. This saves time and creates a better user experience.
In the case of flash on hard drives, Robson can economically support a larger cache than drive makers might consider. Currently, no hybrid drive is being manufactured, although a couple of companies say they expect to ship product by the end of March.
On the other hand, Robson has some downsides for computer vendors. The architecture would add extra cost to PC notebooks in a market that frowns upon extras. And theres currently little perceived value in these caches.
Intels control over the flash chips used for Robson implementations may also be a problem for some. In presentations at Junes Flash Memory Summit conference, Knut Grimsrud, director of storage architecture for Intels storage technologies group, said Robson machines would only use Intel NAND flash for QA (quality assurance) reasons.
Yet for Apple, these points might not be deal breakers. The company is used to including proprietary interfaces or connectors on its hardware, especially for its notebooks. For example, Apples MacBook Pro uses MagSafe, a magnetic power connector that can keep your notebook on the table instead of being pulled down to the floor.
In addition, Intel would be seen as a trustworthy partner for this outside technology. The company is Apples technology partner now. Since Robson takes control of the flash management and will use its own home-grown wear-leveling algorithms for the onboard cache, this might be a plus for Apples concerns about QA.
Certainly, the inclusion of Robson on Apple systems would bring an improved user experience, something that Apple values.
Finally, at the June 2006 conference, Intel representatives kept pointing to June 2007 as the expected arrival date for the Robson technology. Now, that date may have been bandied about because Microsoft at that time was pushing a requirement that mobile systems include a flash cache or hybrid drive in order to receive the highest Vista logo.
But that timing also fits well for Mac systems that ship after the release of Leopard.
In a ironic touch, Apple could become the first system vendor to ship a machine with an expanded flash cache architecture as standard equipment. Currently, no PC vendor has announced support for Robson or for including a hybrid drive.
Of course, neither Apple nor Intel has said anything about Robson on a Mac. Just as we would expect.
However, Intels Grimsrud told me last June that there was nothing to prevent Robson technology from being used by other OSes.
“Theres no commitment yet for other operating system support based on what customers want,” he said.
Sounds like a go to me.
Lessons From WWDC 97
As I mentioned at the top of the story, the 1997 reorganization created a stir inside and outside Apple. It was announced at the Apple shareholders meeting in the first full week of Feb. R&D and marketing organizations were moved around among the top brass.
Then CEO, Gil Amelio, said the company would trim $400 million in R&D expenses and administrative overhead. Analysts expected cuts of 2,000 to 3,000 employees.
At the meeting that attended, Amelio told the gathering of shareholders and employees that Apple “had to get smaller before we can get bigger.”
But how much smaller?
A month later, the other step fell and Apple announced the actual cuts. In addition to the 4,100 person layoffs, Apple killed a long list of technologies and products including Mac OS programming tools, its OpenDoc bento-style component document environment; the CyberDog internet application suite; the Open Transport TCP/IP networking framework; and many others.
Some of these products, such as OpenDoc and CyberDog, were seen as strategic platforms for third-party developers, particularly ISVs.
Apple R&D was mostly gutted as well. The company had long-standing groups researching all kinds of things, including agent technology, mobile devices, interfaces and buses (remember that IEEE 1394 FireWire was started at Apple).
Actually, other cuts occurred later in the year and beyond. For example, this first round left the Newton effort untouched. It was killed in Feb. 1998.
Just a couple of months later in mid-May 1997, Apple held its developer conference and laid out its OS roadmap. The big plan was called Rhapsody, which would include a set of “Yellow Box” APIs derived from NeXT OpenStep that would run on PowerPC and Intel processors.
Developers were pushed towards Rhapsody, even though almost all customers were using the Classic Mac OS. There were many questions and concerns as might be imagined. Developers were mainly worried on all the work that would be needed to rewrite their products and the foreseeable total lack of customers for this new OS.
This Rhapsody Yellow Box plan couldnt hold and a year later, Apple returned to the WWDC with a revamped strategy for Mac OS X. It provided “Carbon” APIs that let Mac programs gain the stability features offered by the Unix-based OS. This improved legacy approach, along with the compatibility “Blue Box” module, pleased Mac developers and customers.
Back in 1997, Steve Jobs wandered to a back room at the conference and held an impromptu “fireside chat” with developers. He was still just the co-founder then; Amelio was still the CEO. Unlike the carefully staged events where hes seen nowadays, this was just him talking off the cuff, unscripted.
During this discussion, Jobs expressed a number of watchwords that the company has more or less held to over the past decade.
For example, he was heckled by a developer who said that Jobs didnt know anything about fixing Apples problems. Jobs responded by admitting that the heckler was probably right. At the same time, he brought the discussion around to customer values.
“Youve got to start with the customer experience and work back to the technology—not the other way around. Ive probably made this mistake more than anybody, and Ive got the scar tissue to prove it,” Jobs said.
Addressing developer concerns about the killing of popular frameworks such as OpenDoc, he said he was sorry for being “one of the people who put a bullet in your technology.”
This harsh talk didnt sit well with some in the audience. But notice that Jobs wasnt sorry about the decision, rather about of being the decider. OpenDoc and CyberDog may be fine technologies but they were to be put down: Bang, bang.
To Jobs, Apple needed to focus, and “focusing is about saying no.”
Besides, he continued, “the rest of the world wasnt going to use OpenDoc anyway, so why should Apple do it?”
Now, you have to smile at that statement, made before a crowd of Apple developers and users in 1997. The rest of the industry was telling Mac users why bother with a computing platform the rest of the world doesnt use.
Still, over the years Apple has expanded its support of industry standards in software and hardware.
According to Jobs at the time, Apple shouldnt be perceived as different; rather, “its important that Apple is perceived as better.”
This last note may be the pitch at the WWDC coming in June. With its Intel transition behind it, Apple hardware is as close as its ever been to its PC competition. And with 50 percent of Mac customers saying theyve switched to Apple hardware and software from Windows XP, Apple seems to have turned around some of the perception part of the equation.
The new question will be, “Can this all continue when Leopard faces up against Windows Vista?”