Progress Report July 2003

tools for developing, distributing, and installing software

By Bram Moolenaar, August 4 2003

OSCON presentation and BoF

The O'Reilly conference was visted by about 1400 people. That is quite a lot, considering that other conferences are not doing well this year. Besides the huge number of presentations there were a lot of opportunities to meet with all kinds of open source software developers.

The Aap presentation went well, although there were not as many attendees as I expected, about 25. There were eleven presentations in parallel and a build tool isn't "sexy" enough to attract a lot of attention.

The BOF sessions, together with SCons, were visited by about ten people, including Guido van Rossum. Not many people, but they were very interested.

Zimbu Award

To stimulate developers to help making Aap and Agide grow and improve the Zimbu award will be handed out. Everybody helping the A-A-P project has a chance to get one of the three prizes.

The award has been announced at the O'Reilly conference and various places on the internet. A few people have already made their first "Zimbu points" by submitting patches for Aap. More promotion is still desired, spread the word!

More info at http://www.a-a-p.org/zimbu_award.html.

Progress

Because of the O'Reilly conference and a week of holidays only a couple of weeks remained for working on Aap. Most of this time was spent on communication, fixing bugs, improving the documentation and small additions. Thus lots of small things.

Compared to the plan for version 1.0 the work lags behind about three weeks. Most of the planned work has been done, there are a few remaining work packages planned. But now that version 1.0 has been released it is about time to adjust the plan.

Coming months

Only two more months are remaining for the work funded by the Stichting NLnet Labs. Let's try to use this time as well as possible. The following is a short overview of what currently is planned to be done. Reactions from (potential) Aap users have guided the choice for what to do and what to drop.

• Followup on the 1.0 release: Cleaning up, bug fixing, improving the documentation, including patches, updating the web site, etc. This will take quite a bit of time, since the code is quite big now and every change must be tested for not breaking existing recipes. This will take about three weeks in total, spread out of the coming months.
• Add an autoconf-like configuration feature. SCons also has a similar feature, it would be a good idea to attempt working on this together. Joerg Beyer also has done some work on this, but has not shown progress the last two months. Closely related to this is automatically finding tools installed on the system (including compilers) and using them for building. I intend to work about two weeks on this.
• Simple installer. Quite a few people apparently find it too difficult to start using Aap and especially Agide. It can be made easier to install Aap by providing scripts that they can execute and/or clear point-by-point instructions. A week will be spend on this.
• Version control. The original plan was to add support for a system that does locking. However, I don't have experience with such a system. Instead I intend to add support for Subversion. This is expected to become very popular soon, since it is much better than CVS. We are just waiting for the 1.0 release. Since Subversion behaves very much like CVS this should take less than a week.
• Recipe finder / package installer. This is working for a limited number of packages. A need for users to upload recipes themselves has not been noticed. Therefore the work on this will be limited to adding a few packages and/or support for specific systems. This will not take more time than a couple of days.
• Dependency checker. Remarks have pointed out that the situation where an included header file is to be generated isn't handled properly. This is an essential feature, it will be implemented and tested. Also, adding another dependency checker should be made easier by moving some of the functionality to generic code. This should take less than a week.

Work packages that will not be done:

• Package management. This has been asked for by several people. But this only makes sense when something really good will be developed, with proper handling of dependencies between packages, reliable uninstall and upgrade support. There are already too many half-solutions. And making a good solution takes a lot of time, it does not fit in the schedule for Aap.
• Console IDE. Despite that Agide provides a nice way to use gdb and Vim together there are not many people that start using it. The dependency on WxPython creates a threshold for starting to use it. People do ask for a debugger window in Vim. Apparently there is more need for a console version of Agide. Unfortunately, implementing this will be a lot of work, too much to fit in the plan for A-A-P. This was never planned to be done anyway.
• Cross referencer. Nothing has been done on this and it has not been asked for. Let's just leave it out.
• Improving Agide. It is easy to think of features that could be added to Agide. But this can be done by anyone. Therefore this will be left to others.

In total the work is just over eight weeks. That doesn't leave room for holidays! Hopefully a few people will be stimulated by the Zimbu award and do part of the work.