Gjallar instance for Gjallar development
We wondered if it would be useful to create a Gjallar instance for managing Gjallar development. We think it would, because the process of creating the instance could organize the development of Gjallar, and helping communication among the developers.
- What would be the artifact we want to create? (an image?)
- What would be the goal of the instance?
- Where to host it?
Göran Krampe: gjallar.se (this box) is a Debian box that can host a Gjallar installation. I just haven't gotten around to it - but now that we do have a working(?) Linux port then we should to it ASAP IMHO.
Update: Göran Krampe: We have one instance up now but it has some issues - I need to configure Apache better for URLs (or get Squeak to skip "/seaside/Gjallar") and we need to figure out why the CSS loads take lots of time.
2006-12-20 Göran Krampe: Another idea is to host this at Toolkit which has hardware just sitting around. :) A third option is to rig something on hardware sitting in my attic doing nothing - but then we need to figure out how to deal with dynamic IPs.
2006-12-20 Herbert König: Remember I offered to hack in the issues from this wiki, if anyone cares for the history.
2006-12-21 Balázs Kósi: Cool! What's really interesting is: how we manage to load those issues in. Lots of questions will pop up! Some of the first ones: Do we start to build a new process from scratch? or from simple prototype? or we pick one of the sample processes (which one?) and start to transform it on demand?
2006-12-21 Herbert König: On the old ones I thought of using Simple and copy paste everything into notes. For new ones I first thought of creating a custom process. Now I think no single process fits everything and a complicated process will be a hindrance.
So I propose we start everything with the simple process and make up a new process if we encounter something that needs discussion, proposed fixes and a decision before being fixed.
2006-12-21 Göran Krampe: Why don't we do it like this (as soon as we get the technical issues fixed):
- Create a new process. No point in using an existing IMHO.
- Use the stages that we currently have on the bug page:
- NEW. Just created, noone has looked at it yet.
- OPEN. Being worked on, hang on. :)
- RESOLVED. Figured out, but fix not yet applied.
- CLOSED. Fixed and closed.
- POSTPONED. Will be fixed, but later.
- Just add the ones we have from the bugs page here manually. Or if we want to make an exercise out of this - write an importer (based on the MigrationService code I cooked up) that can read some simple XML grammar and feed the cases into Gjallar using that format.
- Add a POP3 account for accepting reports by email.
- Add accounts only for us developers and hand out new ones by letting people ask for them.
- Add some option to make Processes available to the user "Anybody" so that non logged in users can view cases. Or?
2006-12-21 Herbert: No objections here. Your proposal is simple enough to not be a hindrance. We can make up new ones as needed. I don't plan to make an exercise out of entering the old ones but will gladly step back if someone wants to :-))
2007-05-09 Keith: Mercurial generates ChangeSets which according to the Mercurial Tutorial can be emailed. Thus Gjallar could keep a record of changesets for generating a software project's repositories. This would be part of a 'software development' process toolkit/component/whatever.
2007-05-09 Keith: I Now have a linux-box sitting there not doing much. This could be a temporary gjallar-dev instance.
2007-05-09 Göran: We actually now have a Hetzner box with plenty of RAM and disk and I have already set up a Gjallar on it (though currently turned off). So that base is covered. But I don't want to spend too much time messing with this during the next 6 weeks. :) But I can be persuaded to do it if we all think it is beneficial - after all, it would help us find bugs etc since we would be running it ourselves.