Updated 2018-04-22 12:16:41 by paul

This page describes the First European Tcl/Tk Users Meeting held June 15-19, 2000 in Hamburg, Germany.

The presentations of this meeting are available at the EuroTcl site.

The notes below are primarily from Jeffrey Hobbs except as otherwise indicated.

I know the title was officially the First European Tcl/Tk User Meeting, but it was so successful I'm calling it a conference anyways.

I'll give some overall thoughts and then go to a session-by-session review of some notable bits. For all that this is a LONG message, a lot went on in that short time which is worth noting (and thus reading, of course :^).

The overall mood of the people there was excellent and you could easily say things went very well overall.

Attendance was officially 65 (65 people paid the DM 100 fee), but more were evident in attendance (I'd say as many as 75 total). The attendees were 2/3 from Germany (conference was in Hamburg), with the last 1/3 being spread around Europe. It's hard to say specifically why the it was so strongly Germany weighted, there were many proferred opinions. The high-tech branch in Germany is clearly one of the strongest in Europe. The conference was in Germany. It was only implicitly made known that the conference was in English (everyone there took that for a given, but someone noted they weren't 100% sure). Outside the newsgroup, the event was only minimally advertised in a German computer magazine (ix). You can add your speculation, but all in all, the attendance was actually much higher than expected, especially given the low-key advertising for it.

So onto the sessions...

The first talk was the Tcl Update, which I gave. It was an abbreviation of the talk that John Ousterhout and I gave at Tcl'2K. The roadmap poll (asking users what they would like to see) turned out somewhat differently than from Tcl'2K. The results were (in votes for each, out of approx. 70 people):
        3        Improve Tcl Performance
        12        Smaller, more modular core
        5        Archive file support (.jar/.zip)
        25        Standard Libraries
        5        Unix binary distributions
        20        Tcl Installer
        25        Standalone executable support
        8        Further Java integration
        12        Drag & Drop
        8        Improved Windows Tk performance
        25        Printing support
        8        Tk abstraction layer (TkGS)
        7        Megawidgets (roll your own)
        35        New standard widgets
        8        Focus on I18N issues
        3        Thread support for Tk

In contrast, the most important item at Tcl'2K was to improve Tcl performance. The issue of themes (skins) came up at one point, in a discussion between Tk and Gtk. While I agree that it looks sexy and would be an optional nicety, I asked how many users would actually use it in app they would ship, and *noone* raised their hands.

In another interesting difference, Unix was an important platform for almost every user at Tcl'Europe, with Windows important to about 40%. At Tcl'2K, it was about 80% Unix and 50% Windows. Each had a few where Mac was also important.

Andreas Kupries followed with a discussion of Stacked Channels. He gave a good background on the needs for it, the history, and the basic structure.

Uwe Zdun continued with a paper on XOTcl. This was a paper that was also presented at Tcl'2K and makes good reading. XOTcl really is an object model based on the Tcl philosophy, and Uwe went through some examples of how it can be used.

Juergen Schoenwaelder, author of Scotty, followed with a paper titled "Married with Tcl". It was similar to Don Libes' paper on writing Expect. It was a very good experience paper that describes maintaining an extension over time, and the love and pain that has to go along with it.

Following that was a CORBA Language Mapping for Tcl paper by Frank Pilhofer. He previously wrote TclMico, know called Combat, which ties Tcl into CORBA as both client and server.

Next in line was Jochen Loewer, with a paper on tDOM, a fast XML/DOM/XPath extension for Tcl. I don't want to go into the technical details, but for those working in similar areas here, this is probably a good paper to read. Keywords DOM, SAX, XPath, Serialization, XML-RPC.

Jeffrey Hobbs followed on with his second talk, this one on Ajuba2. He gave a rough overview, with the breakdown of the parts.

Oliver Schmelzle made a very good presentation on Vignette's work, focusing on StoryServer, and the new V/5. Oliver focused on the architecture of StoryServer, and how they manage the 10s of millions of hits to a site. At one point he noted that Vignette was "having to" move away from Tcl into ASP and JSP due in part to customer demand, but also due to the lack of being able to find Tcl developers.

Following that was a talk by Anselm Lingnau on mobile agents. This is an area where Tcl has been used before. It had some interesting research aspects, but the summarizing point was made at the end where in response to a question he admitted that the killer app for mobile agents hasn't really been found.

The first of two presentations from Patschke & Rasp was given by Ahmet Keskin about a Load Testing app in Tcl/Tk. This was a classic app paper using an interesting and popular area for Tcl - testing. This is a repeat of a Tcl'2K paper again, but a good paper nonetheless.

That wrapped up the first day. A lot of chatting went on during the breaks, lunch and the following dinner. Someone will probably have to come along and shake my brain to get all the tidbits out though. On to the next day...

Friday started off with another app paper presented by Matthias Lüttgert, describing an application for managing the db data associated with waste water management that is actually used by German municipalities. It was essentially a walk-through of the app, with reasoning behind cause and effect for choosing Tcl. It was very classic in how they came to Tcl. A project started in 1991 with C/Motif, and migrated to Tcl in the mid-90s with significant success.

'Jan Nijtmans followed with a discussion of his Wrap application, which creates a standalone Tcl executable. They are working on more elaborate solutions for the future (better obfuscation, ...).

Franco Violi was next with a talk about integrating Tcl/Tk into legacy applications. This didn't seem to have an interesting title, but then I found out that this meant putting a Tcl interpreter into COBOL!! With all the event loop and everything. This was rather clever, and made for an interesting talk.

Carsten gave a quick talk, more of a WiP on ASED (a Tcl/Tk based programmer's editor) and MSGedit, which is a message catalog editor.

After the break, Michael Haschek gave a talk on T-IDE (Tcl IDE), work being done in cooperation with ICEMCFD. This is a SourceNavigator like development environment in and for Tcl.

Following that was a presentation of VisualGYPSY, a Tcl/Tk GUI builder by Patscke & Rasp that has been recently released as open source. It's still supported, and looks like a useful GUI builder, if you need to recommend one to anybody.

Lindsay Marshall (of tclCheck and Frink fame) gave a talk on his tools, which were basically the first well-known static syntax analysis and pretty-printing tools for Tcl. It was a funny talk, and the tools do many interesting things to boot.

Hartmut Schirmacher presented tmk, a Tcl Make tool. There are a series of tools along these lines also noted in the paper. There are always two camps, fairly hardened, on whether one should move away from the ubiquitous, but not totally functional, make, into Makefiles that can be truly scripted.

Following that was a talk by Andreas Otto on a new Tcl compiler (to C code). This has been questioned on the newsgroup because people thought from his fragile use of English that he wasn't quite aware of the intricacies of Tcl and that it just wouldn't work 100%. Well, it was a very good talk and showed a potentially promising compiler based on sound principles. He even plans on making a Java backend. I was convinced that he knows what he's doing and that it could be useful for some. Of course, it's a commercial product...

After that was another talk by Anselm Lingnau, this time on his personal project TkDVI, which is a TeX DVI previewer. It had some cool features, but still needs some work on getting embedded PS and such.

Steffen Werner gave a presentation on XTCC, a tape control system written in Tcl/Tk that worked in very high-end mission critical systems. This was a good example of a successful use of Tcl/Tk in yet another mission critical area.

Andrej Vckovski of Netcetera presented Webshell [1], an app we've seen before at conferences that is reaching v3.0. The interesting new features are that it is thread-safe, to make it a perfect match for the threadable Apache 2.0 release. He noted that Tcl is uniquely poised for Apache 2.0 because it is the only scripting language that has mastered the multi-threading arena to date.

The final paper was from Carsten Zerbst, a discussion of some of the work he has been doing for his PhD thesis. It was an application to deal with handling of data from various sources. It used CORBA and Tcl together. This was paper 2 on CORBA/Tcl integration at the conference, and another alluded to it (Loewer's IIRC). There was definite impetus to push farther with CORBA/Tcl work from several there.

So that was all the primary content. As I said, lots went on in side discussions as well, but this message is long enough already.

On a final note, everyone was quite satisfied with how things went, and there are plans to try and organize something again next year. Carsten will announce later this year whether he will be able to do the organization again, or whether they'll have to look to someone else (like Usenix's European cousin).

I'd also like to thank Carsten for taking the time to organize this event. The stimmung was positive and the outcome very good. It was an impressive and valuable return for the effort that he put in.

Notes from Andreas Kupries (not as detailed as Jeff's)

I am glad that I went to it. I fear that I mostly talked to people I already knew by sight or at least by name. Oh well. It was good to get some more pictures into my memory.

The event wasn't as thoroughly organized as if it had been done by USENIX or similar organization, but given the fact that it was done by a single person it was excellent.
        Thanks, Carsten.

Carsten already said in his posting that there were about 70 people. Officially registered are 65, but from the sight of it it could have been around 80. Expected by Carsten were 20 to 30. We had people from Norway to Greek and Italy, so it was a truly european event. But despite this high geographic variance the actual distribution of nationalities was much more skewed. Meaning ? See the attendance list aggregated by nation below:
        ------------------------------
        USA                 1        Guess who ? (One try only)
        Greek                 1
        Austria                 2
        Italy                 2
        Norway                 2
        Netherlands         3
        <Unknown>         3
        Great Britain         5
        Swiss                 5
        Germany                41
        ------------------------------

We don't know whether the high number of german attendants is due to the fact that the event was held in germany or due to a higher usage of Tcl here.

The talk I liked most was the one from Lindsay Marshall about his 'static testing tools' tclCheck and frink. Actually the inofficial title was rather
        Brace yourself. Two Tcl Testing Tools.

Appropriate, given their concern with bracing in Tcl.

Future is unsure. Carsten is bent on doing his PhD thesis and after that he is not sure about remaining in academia (and thus being able to organize the next event). Given the domination of german(-speaking) attendants most thought it better to have the next event in a different country and see what comes out of that.

A possibility which came up was to ask the Netherland chapter of USENIX to organize it. If that is done it is likely that the next event will be held in Maastricht.

But nothing is fixed yet.