Feb 19 2000.

OOPSLA 1999 trip report, Monday


As you know OOPSLA99 was in Denver. In one word the food did not compare to last year. Early early! Monday morning, well early, if you programming I got up and walked over to the convention center to register and found long lines. Eight odd lines for 2200 people doesn't cut it.

After getting my material I ran into the UIUC group, Don Roberts, John Brant, Joe Yoder, Brian Foote etc. After discussing the Smalltalk world at large I admitted I was going to attended the 'Condensed Java: A 1-Day Tour'. Much heckling ensued. I crept out to attend my lecture and like many Java programmers I was disappointed. {In retrospect I should have saved my money and put it towards a Nikon CoolPix 950}

Much to my surprise at a conference which professionals attended, people who write text books, who make courses, who set standards for computing, why even invent the stuff, here I was treated to a high school 101 Introduction to Java. Remarks like "that's the hard stuff and not covered here" don't make my day. When the instructor discussed how you should nil out references to objects at the end of a routine to help the GC work better it made me lose it. At the time, fortunately for him, I was as far away as possible from the instructor in this very very large room and I'm not sure I could have thrown my book that far.

Happily at the first break I ran across an old friend who will remain nameless to protect his innocence. He too was enduring the material {You are here because?}, then for the rest of the day we sat together at the front and heckled the instructor by asking questions for which we knew there was no positive answer. The day ended early when time was up and with material not covered, instead of staying late I left. My friend told me it got better when they force the instructors to cover the reference material in the course appendix. Sadly I found out that Richard Jones was talking next door about the GC train algorithm. Sigh I should have attended that tutorial instead.

In the evening I attended the tutorial welcome reception, and ate lots of 'ok' food. During this session I got to talk to some of the Disney (Squeak folks) and people from Cincom the new owners of VisualWorks. The Cincom people, the techies, were very happy. It seems from their viewpoint Cincom knows what they are doing, knows how to treat people, and understands the need to make incremental change to the product and bring it to market. In fact one of the developers remarked he was at the conference to work on customer relations, a novel concept coming from his ParcPlace/Digitalk/ObjectShare days. Since Cincom is privately owned the whole issue with attempting to satisfy shareholders is a non-issue and the owner of the company came out to talk to the newly acquired ObjectShare folks and ensured them that Smalltalk was of utmost priority to HIM.

Later at the Cincom booth a sales person informed me they were building up their Website's Smalltalk page and offered to link to my site {Done on the following Monday}. Wow what a change in attitude, remember as pointed out lately in comp.lang.smalltalk, Cincom is NOT PARCPLACE, so remember that {On thursday we learned that ObjectShare was no more too}. For my Squeak presentation to the Victoria Linux Group which is coming up on Dec 14, the Cincom people offered to send me a box of material and VisualWork non-commerical CDs.

The Disney folks of course were up to fascinating things, but more on that later.

At some point in the evening I met Claus Gittinger the inventor of Smalltalk/X. His magnificent product deserves a peek. Later on Friday at a private demonstration to Dr. David Ungar which I attended, Dr. Ungar said it was the most impressive thing he saw all week and the concepts and implementation the most surprising. Claus grinned from ear to ear, as he should since you cannot get higher praise within the VM community.

Claus has a Smalltalk VM which among other things allows you to type in C code for primitives. This C code which you directly enter into a browser window is then run through a handy external C compiler (along of course with references to your Smalltalk instance vars) and linked into the VM exec. Instead of dealing with bytecodes and a small VM where you JIT bytecode at runtime, Smalltalk/X statically compiles the bytecodes to machine code so the VM executable is large and the image file small. Instead of dealing with external C code for DLLs and primitives, Smalltalk/X includes them all in the image. Claus' objective is to have the same environment regardless of the language. So on a method basis we can have C code, or Smalltalk, perhaps someday APL

But that's not all. He's added Java. Yes you can write a method in Java, or subclass a Java class with a Smalltalk class, or load a bunch of Java classes and run them. Again a mixture of Java and Smalltalk code. In review you can mix Smalltalk, Java, and C. Yes this is a true universal VM.

The mind boggles.

Included of course is namespaces and the ability to create packages, and interface to CVS on a method, class, package basis. Lastly it's multi-headed so you can have one image service multiple X terminals, or via a Java applet, support multiple Internet Browsers. Really you must download a demo and try it.

At the end of the reception, Claus and I ran into Ward Cunningham and Richard Gabriel, I then saw John Maloney and went over to discuss a few things with John leaving Claus with Ward.

Later Claus kidded me about leaving him with two fellows that I didn't properly introduce him to and he innocently asked if Ward knew anything about this language called Smalltalk.

In my discussions about some mac plugin problems with Maloney, I got invited to the VM workshop, and in all the group talked to 2AM. Remember the objective here is to talk to people.

See Monday, Tuesday, Wednesday, Thursday or Friday?