Fate vs Bugzilla for feature tracking

[Michael Leupold]

On Wednesday 21 January 2009 23:51:01 Andreas Pakulat wrote:
> On 21.01.09 23:06:47, Michael Leupold wrote:
> > On Wednesday 21 January 2009 22:14:33 Andreas Pakulat wrote:
> > > On 21.01.09 10:07:17, Aaron J. Seigo wrote:
> > > I'm not bound to bugzilla though if there's a better tool to do the
> > > job, especially as it seems that our installation still has quite a few
> > > problems (I'm regularly seeing database connection errors and its
> > > horribly slow often times).
> > I've recently blogged about bugzilla's performance deficiencies and how
> > they can be fixed:
> > http://www.confuego.org/archives/13-Improving-Bugzilla-search-speeds.html
> Was that also on our planet? I recall I've seen some bugzilla related
> blog (not Aaron's) but decided to not read.

Yeah, that was it. I didn't post the full blog but linked it because it was 
rather long and would have taken too much space.

> > I already prepared a patch to Bugzilla to allow others to integrate what
> > I developed and submitted it for their review. If it gets accepted I'd
> > also backport it to make integration with bugs.kde.org possible. I can
> > currently think of two drawbacks my solution has:
> > - The indexes are huge (I'd guess around 40-80 GiB on bugs.kde.org)
> > - New changes aren't pushed immediately to the index. Multi-level indexes
> > ("deltas") are cheap though.
> Hmm, maybe I should've dived a bit more into the text-search stuff at
> university when studying db-technologies :)

Well, not too hard actually. Sphinx does most of the stuff itself so you don't 
have to bother :-)

> Anyway, just wanted to add that most of the time its slow for me when
> loading a bugreport or sending a change to the server. Of course that
> could just be somebody else doing a larger query on the db stealing time
> from me :)

Yes, that's quite possible. Judging from my local installation for testing 
there might be several causes:
1) People doing queries that are dog-slow and take forever
2) Pushing long buglists into the list-template

The first one was the one I set out to solve, the second one should be a non-
issue on bugs.kde.org as the number of bugs returned is limited (afaik).

Regards,
Michael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20090122/81b7160f/attachment.sig>