April 91 - News from Phoenix
News from Phoenix
Now I know how my final, outgoing column as editor in the last FrameWorks should have ended: old editors never die; they just come back and live on forever as columnists!
Well, forever's a long time, but I'm certainly here for the near term, and definitely in a new capacity: basically, news and good general gossip seem to be the prescription.
Everybody's just back from an exciting week in Phoenix, Arizona, and there was certainly enough going on there to fill up a few pages with news and gossip. Let's jump right in.
Apple UnvEIls MacApp 3.0
TO A LARGE AND ENTHUSIASTIC CROWD
I and most of those who showed up (over 220, just about double last year's attendance) got our first look at MacApp 3.0, and it looks good. Steve Friedrich, Lonnie Millett, and a much expanded MacApp engineering crew from Apple have been busy-there are a couple of pretty major changes in the works, to say the very least.
3.0 Dependency Change Mechanism
I think the niftiest new feature of MacApp 3.0 is the dependency change mechanism. This mechanism should be familiar to any Smalltalk aficionado. In simplified terms, DCM lets objects in one part of your program automatically inform others when they've changed, so that those dependent objects can then take appropriate action, whatever that might be.
The MacApp team came up with a particularly graphic example to demonstrate a good use of DCM: before-and-after versions of the Calc example. One implementation shows a recalculation of Pascal's Triangle without DCM, the other with DCM. The traditional "without" version took about 20 seconds for a complete recalculation, the "with" version was nearly instantaneous.
The demonstration was neat! Apple engineering types obviously thought so too, and showed it off in at least three different sessions during the week.
MacApp 3.0 Written in C++!
This was the bombshell at the Conference: Steve Friedrich's announcement on day one that MacApp had been totally redone in C++, and that source for 3.0 would only be available in that language and not Object Pascal. Boy, did that get a reaction! This was the most divisive and political issue I've seen come up in over four years of informative and enjoyable meetings between developers and Apple. Talk about some major differences in opinion! This was the main topic of conversation throughout the week.
Friedrich announced that the MacApp team had taken several weeks and converted the whole works over. Which was great news for C++ users, of course, but the intimation was that starting with 3.0, an Object Pascal version of MacApp would no longer be supported. In particular, THINK Pascal users would shortly be unable to use the wonderful debugging features that they were so happily used to. (Note that compiled MPW Pascal code would still certainly be linkable; however, there'd be no way to link with THINK code-at least at present.)
Dave Neal looked a bit shell-shocked, other THINK Pascal developers at the Conference looked none too happy, and Raines Cohen of MacWeek, with his Portable slung off one shoulder, gathered quotable quotes from everybody in sight.
What was left unclear, after all the dust settled, was the question of just how inevitable is the changeover: Friedrich and his engineering crew seemed to think it was a done deal, but product management didn't seem quite so sure. The reaction from the Pascal community was very, very strong, and Apple had to publicly back off a bit. They are now taking some time to reconsider the issues, or at least to be a bit more thorough in their considerations before lowering the boom once and for all.
Tom Chavez told me that Apple's put the question out to an independent survey company, which is talking to a good cross-section of MacApp developers in order to get better information on which to base an informed decision, and that the issue should be settled one way or another within a month or so.
My own hunch is that C++ is on its way. I suspect that it probably is a done deal, and that like it or not, Bjarne Stroustrup is going to get a few new users. As they said lovingly in the movie, Poltergeist: It's here!
MacApp App Debugs MacApp
Apple made a couple of System 7.0-related product announcements at the Conference. I thought that SourceBug-an MPW-compatible, double-clickable, standalone debugger-looked particularly nice, and I'm looking forward to getting a copy. It's one of the first Apple development tools I've seen that retains a lot of the nice and easy, casual feeling of the Symantec environment.
Dave Wilson asked me to make sure that Mike Lockwood got credit for writing the app, and to point out that he did it in MacApp.
Lighter moments too
Bjarne Stroustrup and his favorite language, C++, provided some of the lighter moments at the Conference. The Language Wars panel on Wednesday night was a big success, with a large audience, and five panelists all happily cutting each other to shreds in front of a highly partisan crowd for close to two hours.
My only pre-session instruction to the panel was that all logic was to be left at the door, and indeed it was, except for Dave Wilson, who got there late and actually tried to talk rationally and make some sense. Boy, was he out of place!
Yes, there was more, but you had to be there (and you will be next year, won't you?). And with that, I think I'll call it a wrap. Take care.