Well, everyone's home from Apple's 1992 Worldwide Developers Conference. The main theme of the week was that they're doing their best to get new technologies (hardware and software) out the door in greater numbers than ever seen before, and developers will have to work hard to keep up.
Some things to remember:
The main focus of the afternoon was on AppleScript. Apple has been telling developers to incorporate Apple Events into their programs for a year; now they're really showing us why. The AppleScript language is due to arrive this fall. With it, you'll be able to script and record Apple Event-driven applications. Also, the Finder will finally support real Apple Events. AppleScript compiles when saved and decompiles when opened, so you can write a script in an English dialect, save it to disk, ship it elsewhere, and have it open in the Kanji dialect.
On Monday afternoon, Apple also announced plans for the cross platform framework (more on that later), a new linker, and new compilers. They're talking about a native C++ compiler and an incremental linker that will cut link times from eight minutes on an FX to one minute.
Tuesday night's session was on Writing the Ultimate Macintosh Game. Speakers included Jim Reekes, who talked about the next Sound Manager, and Bruce Tognazzini, who, of course, spoke on user interface issues including copy protection methods and making assumptions.
The new printing architecture is also impressive, making printing as easy as it's supposed to be. Users will be able to have desktop printers and use drag and drop printing. Serial computers will become sharable via file sharing. Print monitor will disappear, and we'll have true background printing for all printers, not just LaserWriters. Printer drivers will be easier to write, as well.
There were several other interesting sessions on Wednesday. They included a session on Apple's new cross platform development, where QuickTime for Windows was demonstrated. It was impressive, until it crashed. Apple seems to be committed to the idea of providing their technologies on other platforms as users and developers demand it, so tell them what you want and where you want it.
The coming translation manager was also demonstrated Wednesday. The manager doesn't do any translation itself; it provides a good user interface to translation systems such as Claris's XTND. Besides allowing users to open documents whose applications aren't available without ever seeing the "application could not be found" dialog if there's any application on the system that the document could be translated to, the translation manager provides translation of material on the clipboard. When you copy from Microsoft Word and paste into macerate, all formatting will still be there. The same applies to editions. One side benefit: the translation manager has to know more information about a document than the Finder has provided, so Finder lists will say "Microsoft Excel Chart" instead of just "Microsoft Excel document."
Finally, we saw both Caspar (which has gotten a lot of press coverage lately) and the MacinTalk replacement. Caspar really does have continuous speech recognition for limited domains. The new speech-to-text generator is fairly impressive as well. It comes in two versions: the low quality version takes about 100K of memory; the higher quality takes around 1.5 Meg. Pronunciation of abbreviations is influenced by context, and punctuation affects inflection.
Later sessions on Thursday described WorldScript, Apple's implementation of Unicode, which seems to be quite clean. It will provide system-level support for Hebrew, Arabic, English, and Kanji in a single document. It seemed to impress even the Japanese and Taiwanese attendees. WorldScript is to be the major feature of System 7 Release 7.1, which Apple hopes will see a "golden master" seeding this fall. System 7.1 will then serve as the basis for OCE, AppleScript, and QuickDraw GX.
Several of the afternoon sessions were on PowerPC. The slogan "The Future of the Mac is PowerPC" was repeated just often enough to get annoying. One of the more interesting items on the agenda was a discussion of FlashPort which (with some hints) could translate the binary code of a compiled Mac 680x0 application directly to the equivalent binary for the PowerPC. Estimated time for translation is 2 days to 3 weeks. Compare that to a port. Unfortunately, the FlashPort requires an RS6000.
Looking farther into the future, Apple demonstrated a very early prototype of pen computing. The big focus for pen computing seems to be "the pen is more than a mouse." Pen-ignorant applications will be able to accept printed input because of recognizers built in to the system. Pen-aware applications will be able to give the recognizers hints. Sounds exciting.
Thursday night was open, and all good MADA members attended an exciting Bamada meeting (see article in this issue). Fortunately, we all got there despite the charter bus driver not knowing how to get to the De Anza 3 Auditorium at Apple.
Dick Trismen (of MADA fame) demonstrated a sample application that uses a pre-release version of Bento. In brief, he used Bento to add support for embedded objects to the Edition Manager. This was a pretty neat trick, although he lost about two-thirds of his audience when he went into a blow-by-blow account of how he did it. The upshot of the demo was that Bento is easy to use, and that there is no reason not to use it-except, of course, that it's not available. Apple hopes to publish the Bento 1.0 spec this summer and hopes to give the rights to the spec to an independent organization it hopes to form, which it hopes will be called the "Bento Alliance," which Apple hopes will help make Bento a platform-independent document standard. We should all hope they succeed.
Looking farther into the future, Kurt Piersol introduced Exemplar, Apple's proposed application architecture. The basic idea behind Exemplar seems to be to take on object-oriented approach to applications. In a data-centric world, one focuses not so much on the application as on the tool that manipulates the data. Now these are the same, but in the Exemplary world, one would install a "PICT tool" that would be used to define, edit, and render PICTs in all applications. Likewise, one would install a "rich text tool" that would handle all text editing, formatting, and display. Then, one could combine PICT and text objects in an application which knew nothing about either PICTs or text. The application would just tell the data's tool, "this PICT has been selected," or "there's been a mouse-down in this text, at this location." All the application would need to know would be which tool was associated with which data. This future may be one reason that Apple keeps telling developers to write applications which "do one thing and do it well."
Friday afternoon featured a Stump the Experts game show in which the audience and panel tied, a follow-up to last year's debugging session (the best tool featured here was SmartFriends), and Q&A with Apple management.
