[Nepomuk] Some conclusions/ideas after 2 weeks with the project

Paweł Paprota ppawel at fastmail.fm
Mon Jun 20 12:12:28 CEST 2011


Hello,

I've been digging into the Nepomuk project for the last two weeks. I 
started with the idea of using Nepomuk as a basis for a media library 
component - kind of a semantic representation of user's movies/TV shows, 
music etc. Quite similar to what XBMC project is doing but with Nepomuk 
underneath and KDE (possibly Dolphin?) as a user facing UI.

However, right now I feel like this work should be postponed because of 
the problems I'm having with the project itself. For example right now 
Nepomuk is not able to index my home directory. One of the causes is:

https://bugs.kde.org/show_bug.cgi?id=276056

Sometimes nepomukindexer crashes (I have a core dump indicating there is 
an assertion failure in Strigi) and indexing hangs. Also I had to 
exclude *.pdf from being indexed because of:

https://bugs.kde.org/show_bug.cgi?id=274895

I see a lot of open crash level reports in Bugzilla. I don't feel I 
should proceed with adding new functionality (media library) before 
stability issues are somehow resolved and there is a regression test 
suite for this kind of issues.

I thought about preparing a continuous integration environment for 
Nepomuk so crash/load/unit tests could be run constantly or at least 
after each commit. This is a normal thing for me to do in commercial 
projects. Especially with this kind of component there should be stress 
tests running all the time - otherwise it's the users who will encounter 
crashes/hangs etc. and indeed they do as is represented by state of 
Bugzilla reports.

Sadly I've just cancelled my Linode account few weeks ago because I 
haven't had the time to use my virtual machines - otherwise we could use 
my account to launch CI server and run tests there.

However, I'm sure the infrastructure can be set up other ways, in the 
end I could simply create (and pay...) for new Linode account.

First I would like to know your opinion about what I should do right now 
because I am torn. In my commercial projects I can't stand it when there 
are open bugs at the level above major, especially crashers and 
blockers. I would love to be able to further familiarize myself with 
Nepomuk by triaging existing bugs and trying to track down, reproduce 
and test crash/blocking bugs.

In the meantime I could start writing about media library ideas on my 
blog so that KDE usability folks can get involved as well.

What do you think?

-- 
Paweł


More information about the Nepomuk mailing list