July 19th 2000.	Camp Smalltalk (July 17-20th 2000)

? What is this Camp Smalltalk ?

Camp Smalltalk Annapolis Monday July 17th, 2000 (QKS and ".net" information)

Camp Smalltalk Annapolis Tuesday July 18th, 2000

Wednesday July 19, 2000, CS2

It's morning and Apple is busy announcing neat things and I'm in the wrong city to see it. However I'm still having fun and sipping coffee while watching the GLORP team who are busy looking at the Berkeley DB library. Various pieces from that technology are being looked at to assist with getting data in and out of Smalltalk. .

Bijan is busy working on the XML, still, a forever task he claims, but progress was being made.

Over at the SOAP round table six people in the SOAP team were sitting around, and actually coding by this time in the morning.

The ANSI team was busy setting up the infrastructure for VisualAge to accept additional ANSI standards. An attempt had been made to port some code from VA 5.5 to 4.0, but that failed due to version issues, and movement was afoot to migrate in the other direction. Progress was slow. One of the problems they were finding of course was things like scaled decimals weren't ANSI compatible across all platforms. There wasn't a common dialect to use to ensure the code could port to all platforms. Movement between dialects was messy, mechanical mostly but messy. It seems that SIF hasn't been really accepted by the vendors, and more work needs to be done on the XML exchange system. As mentioned yesterday Oasis seems an interesting choice, later in the day another obvious theory was raised and acted upon.

TeamWork…

Well it's not a group, but interesting to watch. Sames had come over to the GLORP group to make some changes to the SUnits code to enable it to catch halts to enable debugging when running the test Suite. Various other experts flowed around and assisted groups when they encountered difficult problems. Where else can one come and see world class experts at work? Camp Smalltalk of course.

Ralph and Tracy were busy moving test SUnits between two macintoshs. Ralph earlier explained although his ancient powerbook was obsolete it was twice as fast as his old one and only cost him $400.

Your scribe spent most of the morning working on the ANSI test SUnit with Sames. Did you know VisualWorks violates the ANSI standard for returning the first element of an Interval; the class of the object returned for first versus (at: 1) can be different!

Lunch came very quickly, which was a mountain of subway sandwiches paid for by IBM. Thanks IBM.

David Simmons dropped in and was tackled by the SOAP team. David had some stuff from the conference last week on his machine that was being examined to clarify that "SOAP on a rope" spec. Later in the day a Microsoft employee emailed us and offered some assistance, apparently we got linked to from a msdn.microsoft.com web page! I wonder what my ISP thinks, we suddenly became a source of information about SOAP, this was discovered when we were searching for SOAP information with Microsoft and turn up a link to my Camp Smalltalk Story. Amazing.

Remember the issue of moving Smalltalk code between dialects, Peter over in the ANSI group was finishing up a transfer script to move fileouts from VisualWorks to VisualAge. PEPE is the name, based on the notion that all the dialects understand Smalltalk. So why not use doIt code? Implementers need to write four methods on the class side of PEPEImage class that allows you to instantiated a class in your dialect of Smalltalk. Then just add methods, values etc by doIts. It's the simplest thing that could work.

&Mac183; A link to it is here *

The GLORP team was looking at how PostGreSQL returned values and how they were going to translate rows into Smalltalk, somewhat below the level of instantiating the objects. Beside the network infrastructure team was working on some implementations of MIME and IMAP. They decided to merge Les' IMAP client with the Swazoo framework and a MIME framework. This provides a better framework for Swazoo, so for example a http server can consolidate information from lots of sources and act both as a client and a server. Thus you have a rich framework to enable your to access information.

The IMAP stuff was being changed as part of the first pass to consolidate the two clients. Most of the basic commands are there, for example to search the Squeak mail list mailbox, get hits back, move messages between mailboxes.

Ah Craig Latta appeared and I spent a few minutes bring him upto date with where Camp was going and where he could contribute.

Back at the ANSI group some giddy Smalltalkers were talking about Squeak, well mostly, but really they were working on the FloatCharacterizationANSITest. I think they were trying to distract themselves. The FloatCharacterizationANSITest referred to a rather dry ANSI thesis on numerics. Did you know the ISO/IEC 10967 specification is $75. Not a single copy on hand. But Visualworks doesn't implement the standard so another failure is thrown. Again a very mechanical process to make these thousands of test cases to check for ANSI compatibility.

Early in the day we were discussing that the ANSI test isn't a test for a Smalltalk VM programmer to use to see if their Smallltalk implementation works, rather it is a test to see if the Smalltalk is compatibility between dialects. However in my opinion two existing Smalltalks can be ANSI compatible, but code will not transfer due to the undefined behavior in the specification. This is unfortunate and it is up to our community to stretch the specification over the actual implementation we use today.

Over by the XML team Ralph was discussing now the Squeak World Tour is going to work. Bijan was working with a colorized Squeak browser exchanging XML classes with Rodger who was using a B/W VisualWorks editor, quite a contrast. Business versus Artistic. As I mentioned the disjoint, Bijan discussed his culture shock with migrating to VisualWorks, and the missing Squeak features like version on a method level. Mind of course Envy gets you this but it's expensive and more complicated.

The Swazoo team was moving Swazoo to VisualAge, issues with Socket portability was clouding the porting. A movement was afoot to make a portability layer to make things easier for people wanting to port Swazoo.

Back to the ANSI team, Donald MacQueen was juggling nerf balls, and explaining how to juggle. This is a complicated process and who says Camp can't be fun. Paul and another camper were using a colorized browser in VisualWorks (yes it can be done) to build a test suite on Read/WriteFileStream. Progress all around. By now 28 smalltalkers around the room were running full tilt pumping out thousands* of function points per hour. (* your estimates might differ but serious work was afoot)

As I was reviewing these notes and responding to email from people around the world asking for further information about Camp Smalltalk and what was happening I had to reflect what a neat experience this all was, but hey back to work, an empty seat lurks over by the ANSI group, or beside the Network infrastructure group, which way should I jump?

Over the ANSI group and an afternoon of code SUnits for Abstract collection factories.

5:45 rolls about. We find out that the room is rented for the evening, the hotel didn't understand we really wanted the room for 24 hours at a time. Mad scramble and in 10 minutes all the cables, equipment and people hustled out of the room. Amazing what can happen when a room of self starters are given a problem to solve.

Later that night most of us departed for downtown Annapolis to see the sights and to have dinner, some like your scribe did a short tour of the Naval Academy. Then dinner at a wonderful pub by the water in Annapolis to discuss the finer points of life from family history, poly-science, to: Will a version of Digitalk be resurrected for ".net" ? Sadly the evening hours stretch onwards and all good things come to a close and we returned to the Hotel, in driving rain with some quite lost Smalltalkers attempting to give the driver instructions.

But before bed, Bijan, Ralph, Rodger, Craig, and your scribe got caught up in a deep discussion about metaclasses {http://netjam.org/smalltalk/objectMetaphysics} and how to teach that concept to the mass, and if of course can you really need to teach that concept. Where else but at Camp Smalltalk can you come and have such discussions in a hotel lobby. It's early 10:45 and time to prepare a few loose ends, for tomorrow CS2 ends…