On Fri, 2008-01-18 at 17:21 +0100, Murray Cumming wrote: > On Fri, 2008-01-18 at 13:17 +0100, Patryk Zawadzki wrote: > > I mean proper RDBMS. Once you get that in place you can generate > > DOAP/RDF/XML/whatever on the fly with little to no effort. > > That's a nice idea. But it's easier for our developers to enter > information into a file in svn than into a RDBMS. Even if there was a > wonderful web interface, having it in svn makes it more likely that each > svn module has up-to-date information. These are our habits and it's > hard to change them.
Some of the folks here have probably heard me talk about Pulse, the project tracker I've been working on. Pulse crawls around various information sources, sticks stuff into a database, and regurgitates the information back on pretty web pages. I think there's a lot of value in having a centralized store of information that can tell you everything, and that's why I'm working on Pulse. But if you have a database somewhere you have to update, as well as some files somewhere else that you ought to update, inevitably something will be out of date. And that's why Pulse is strictly a read-only system to the outside world. I'm managing to collect quite a bit of data just by analyzing what files we do have, but the information is not always complete or entirely correct. My hope is that it will be useful enough for people to use regularly. Then when they see information that's not quite right, they'll fix the source files. Obviously, having more structured information helps immensely in Pulse. But I don't want to see us using multiple redundant files for things. So for instance, if we specify the maintainers in both a DOAP file and in the MAINTAINERS file, one of them is bound to get out of date. -- Shaun _______________________________________________ desktop-devel-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/desktop-devel-list
