[Kolab-devel] Refactoring and enhancing pykolab (and the Kolab Wiki)
Paul Boddie <[email protected]>
| Newsgroups | gmane.comp.kde.devel.kolab |
|---|---|
| Message-ID | <[email protected]> |
Hello again, I've been taking a further look at pykolab after doing some initial trivial work on licensing boilerplate [1] and style-related changes [2], and I've managed to make some fairly extensive changes without completely breaking pykolab, and these have now been added to the following bug: https://issues.kolab.org/show_bug.cgi?id=2660 Naturally, I managed to break and unbreak things on the way, and the patches attached to the above bug mostly keep things unbroken, but it can be difficult to ensure this especially when submitting "cleaned up" patches. Nevertheless, I've attempted to tidy up the command/component framework underlying the kolab and setup-kolab commands. The apparent need for a few other things arose when doing this work, and so I'll list them below; they all concern setup-kolab... Until now, the command has worked in "do everything" mode by default, which is perhaps a little too enthusiastic, and I'd like to think that it would be more intuitive if the interface presented the configurable components by default (the "help" command) and if it could also show components that need configuring (kind of emulating the service command) albeit with some extra work described below. The command complains very quickly about things like directory server instances and tells you that /etc/dirsrv and /var/lib/dirsrv must be "clean", but it would be nice if it could offer a more concrete remedy or at least hinted at the tools - other than rm ;-) - that could be used. (The Debian dirsrv init script is also unhelpful and needs repeated "prodding" so that slapd actually restarts, and I wonder if we might not try and get this fixed in Debian.) There is also some kind of uncertainty about output and logging. I experienced abrupt exits of the command where something might have been logged somewhere, but on the console there was no indication of failure. Some consistency here would be desirable. I tried to add some "work detection" to the setup commands so that they can tell if they probably don't need to do anything, and this has mostly been done in the database-related tasks. I made it possible to override the behaviour to reconfigure things if this is really desired. Maybe it would be good to perform more in-depth tests of existing things to ensure that they really will support Kolab and aren't incomplete in some way. I noticed that there is a kolab-conf command, but I'm not really clear on what it is supposed to do. If it is meant for testing the configuration, perhaps the above work could be integrated with it, but maybe someone can explain what this command should be doing. On another matter, I brought up the topic of the wiki on IRC, and it was indicated that a new wiki instance might be made for Kolab 3, particularly since the current instance mostly concerns previous versions (and is gradually accumulating account spam). Although I'm sure the emphasis is generally on getting things onto docs.kolab.org as quickly as possible, I think that being able to formulate things on a wiki might be useful for people like me who are trying to be more familiar with the architecture of the solution, with such informal notes and ideas obviously not fitting into the proper documentation. Anyway, I'd like to think that there is potential for making pykolab and the associated tools a bit nicer, so I hope this work is of interest. Paul [1] https://issues.kolab.org/show_bug.cgi?id=2552 [2] https://issues.kolab.org/show_bug.cgi?id=2554