July 18th 2000.	Camp Smalltalk (July 17-20th 2000)

Camp Smalltalk Annapolis Monday July 17th, 2000

Camp Smalltalk Annapolis Tuesday July 18th, 2000

CS2 Tueday AM. It's early on the west coast 6:30 PST, very early if you went to bed at 4:30 EST. Do campers ever get to sleep? After some sweet buns and muffins, which were much better than the hotel food at CS1 we were off to the races.

Today was a day for work, so the report has fewer words, but some of the topics discussed at lunch have great impact for the Squeak community so read on.

Arg no digital cameras around, where are they? Are all those early adopters really west coast people, don't they sell those things here? Enoch's famous expresso machine was in full swing

Stopping by the ANSI group.

The squenceableCollection was being tested, most of yesterday was spent getting people up to speed, hopefully all the collections should be done for four dialects today. (Dolphin, VA, VW, Squeak). I spent a few minutes listening to Ralph Johnson, he was talking about the OOPSLA poster session "Building the ANSI test suite" and how that was a vehicle to get interest in Squeak and in SUnits with the OOPSLA mainstream community.

Over in the corner Donald MacQueen was batting 0 for 1, but that's testing at this time of the morning. The IdentityDictionaryANSI test was failing with doesNotUnderstand. A method existed in abstract collection helper that was missing in abstract dictionary helper. Then it was discovered that DictionaryANSItest was in the wrong hierarchy, well maybe not. Mmm oops they where missing just a little clue, some more effort and a suggestion from your scribe rectified the problem, the tests were in the wrong class they need to be in the helper object versus the ANSI test object..

Later I found them trying to solve the test for collect: transformation on a Dictionary, but according to the specification it should return a Dictionary, of course it really returns an OrderedCollection so the Campers were mystified about the actual intent of the ANSI specification. Later Sames and I found a issue where add: to an collection has an undefined return value, versus of course returning the object added.

Lots of red failure bars up on different screens.

Sames was telling me they were working on the extensional collection abstract protocol. Ralph was pleased and they were well on track to meet their goal of getting all this to work by the end of the day. So what for tomorrow? But before I found out Sames was called over by Donald to solve a problem with resolving the order of the helper object. It seems resolving the method to use is done in reverse order of the collection.

SOAP

Five people were crowned around a laptop. Trying to understand the standard and how to build the software. Various standards documents, vendor material, CRC cards flowed around in a turbulent pile. The issue was the spec was too vague and they were attempting to understand the specification from examples. I suggested they email the authors of the spec. Later they crowned around a Mac G3 and were reviewing the results of a search on the internet for SOAP implementations and behaviors.

Mark was busy writing a requirements document based on input from the campers about how they thought the standard worked, with grumbling about how the spec is so vague about important issues.

Today they * think * they understand what they are going to do… Gee it should be simpler that it appears, but it's not? Or is it? Something about it's a soap on a rope spec, just when you get a handle on it, it slips away.

One table over Bijan Parsia spent a few minutes talking to me about Flow and his testing of it on Windows, then issues about Squeak Socket support. Although this isn't directly a CS2 issue it was of interest to me. Well this was going on from time to time I was trading notes with Rob Withers about his testing of my Squeak Socket Code SUnits suite.

GLORP.

Progress was being made against adding transactions support to PostSQL, another camper was working getting mySQL to run the 106 SUnit tests against that database. They have a special version of SUnit that keeps instances of TestFixtures across a set of tests for an extended period of time. Movement was afoot to bring this feature to the normal SUnit stream. A question was raised about porting Flow to VisualWorks, this would assist the GLORP and the Swazoo people.

Linux. Well not that there is a linux group, rather an individual or two managing the Linux server and the network. There had been a new card installed in the 486 machine for 10Base-T support and both a 100Mb and 10Mb network being reconfigured. Alas the work was not completed, there was a faulty 10Mb card, and we really needed a 10/100 hub to link things together. Ralph tells me that at the next Camp more effort will be put into network infrastructure before Camp starts.

Oh and quite a few of those new USB Microsoft mice with glowing red bottoms were in use around the room.

Lunch ***************

Ernst & Young gave a wonderful lunch.

Ralph got up and talked. Thanked Ernst & Young for their gift. He then went into the background of how Camp Smalltalk started and that we should have conference hosted by STIC that we of course would manage.

{Note of course Ralph is looking for people to help for both ideas below}

We need somewhere for Business to come to so they can connect to people, developers, vendors etc in a conference setting. We reviewed the why, the issue of how much it is should cost, as we know Sigs Solutions hosted a very expensive conference in a very expensive venue, their target was the Wall Street crowd, and once critical mass was lost they dropped it.

Now remember most of the people that are here are self taught, and the average number of books read by a programmer in the USA is 0.4 (Zero . 4 ) Ralph remarked based on the number of books he reads each year then there were lots of people that read zero books per year! So to solve this problem there is a tendency for business to sent people on conferences because the cost benefit of the training they get is very good.

Question: What would a conference do?

A vendor answered this. The loss of a conference is another nail in Smalltalk's coffin, no magazine, no conference, no developers/people. At OOSPLA there is no benefit for us to attend, all the interested people are the same from year to year, but it's a negative not to attend.

Another issue, a conference is a training vehicle, and a vehicle that is used in the USA to find jobs. It was pointed out some of this wasn't workable in Europe.

So why don't the vendors organize a conference. The issue is that people who don't use the vendors product won't attend.

Sames pointed out the community is dead! Why didn't the hundreds of Smalltalkers in the surrounding area come here, lots of bitter truth in this. The number of people needed is too small. This is really like a closed meeting, it's not open. Conferences are viewed differently by corporations, many developers attempted to come but permission wasn't granted because this isn't a conference with an agenda, speakers etc etc.

Ralph pointed out the idea of Camp Smalltalk was to solve some particular issues. It can't solve all the issues in the community.

Sames pointed out that many here could speak at conferences. Ralph was back to the main point that he needs to have some people help with organizing a conference. It was pointed out what's the objective?

Your scribe pointed out perhaps David Simmons could or has some Microsoft funding for ".net" solutions that we could use to promote the conference.

What is the objective?
Well, say 200+ people, a nice location, some media interested, a Microsoft tie in.

Sames stated: Having a conference will double the size of the community!

Ralph said it's our better technology, the message is how to be rewarded for using Smalltalk. Flourishing requires growing, the community is too small.

Someone from Cincom pointed out their options were graveyarding the product, or spending money on the product. But growing the product means growing the community, they DO want to grow the product. Hence things like hosting this Camp.

A conference should strengthen the community, and grow the community. One conference won't do it, many will. For example Java One helps new Java programmers, we don't have that venue. It also helps third party vendors get access to people. Not having a conference makes us look weak.

Lastly it was pointed out that we don't want the objective to be to make money.

Ralph's second idea.

Stable Squeak World Tour

Oops not "Stable" Ok we all laughed, gee bring it up to stable eh?
Perhaps Squeak World Tour.

How to make Squeak more like a business tool. For Disney Squeak is a research tool, not a business tool. Ralph, John Sarkela and David Thomas were talking and David said Squeak was the best way to get Smalltalk out there. David said as long as Disney is driving the process this it won't work. So a World Tour is need to fix the problems.

John Sarkela as the tour guide! Perhaps I too can learn how to make handlebar moustaches

The objective is to clean up the stuff in the Squeak image, there is lots there but it’s a finite task. Ralph also talked to Dan Ingalls about this topic and got approval. We need to refactor, build test cases with the objective to build a more stable platform with a busness flavor, let alone how to tie to ".net" as your scribe points out.

Someone pointed out Disney was looking for a person to do this task on comp.lang.smalltalk a few months back.

Squeak has lots of advantages.

Gee it's a lot of work? Squeak, Camp Smalltalk, Smalltalk conference.

Ok, remember we've asked Disney for feedback on this and we think we do this without forking Squeak. The objective is to get 6-10 people together every other week and hack Squeak for a week. In 8 to 10 weeks we suspect that all the work could be done.

*A link to the World Tour Page should go here *

--------

Lunch weigh heavy on our digestion systems, 24 campers are hard at work, others are also digesting thoughts from the discussion from lunch.

Les Tyrrell was showing your scribe some of use the work he was doing with the Refactory Browser and his Oasis system, a system he has spent many years writing.

http://oasis.canis.uiuc.edu:8080/oasis

Oasis is a system to partition an image so you can bring in other code and not have it destroy your existing image. It's a sandbox solution to strongly isolate code. Execution of code should happen only on MY command, a filein violates this since it really is a program you are running. Although the community hasn't had a problem with viruses it still is a pain if you file something it to find it has changes some base classes in an interesting manner. Les tells me that originally it was a container to bring stuff in and examine in a safe manner, but now is progressing towards the viewpoint of being able to integrate the code into the working environment.

Les gave me an example of filing in the Flow work into VisualWorks. Once in Oasis it clearly show which objects were referred to outside the module, from this point you could rebind the objects to objects in other namesSpaces. For example Flow adds a test isEOL on Character which of course would be unbound. This need to be rebound to the VisualWorks class Character in order for it to work. But given the years of work he has done there is lots of richness in the tools to explore the filed in code at the meta level.
Definitely worth a look.

Later other people were thinking about this and pointed out you can file most of the dialects of Smalltalk into Oasis, then file out to a particular dialect. Think storage!

Your Scribe then spent a few hours working with Sames on the ANSI OrderedCollection Test SUnit. Mark and David where working on PostSQL driver problems and refactored some SUnits for GLORP to make it less Oracle specific, and more generic.

A Camper or two managed to escape for a few hours and see downtown Annapolis. Lots of Ben & Jerries Ice Cream, way too much for some 8 year olds.

Dinner at Chilies and back, just a few Smalltalkers crowded around a G3 and flat screen asking a million questions or two about Squeak and how plugin works, and how segment loading work. Afterwards a few Squeakers crowded around Rodger and talked about numerous issues, like the problem with SUnit and Morphic and then realized that Rodger had fixed all these problems, so changes sets flowed across the wires.

Buried in email, it's fast approaching 11:00pm and campers are busy solving the worlds problems, or at least issues close to their hearts and I'm retiring for the evening.

Tomorrow…