[Nepomuk] indexing errors in Nepomuk output, indexing crashes

[Sebastian Trüg]

On 12/02/2011 02:02 AM, S Page wrote:
> On Wed, Nov 30, 2011 at 09:01, Sebastian Trüg <trueg at kde.org> wrote:
> 
>>> (filewatch service) KInotify::slotEvent: "/." EventIgnored
>>>         ^^ another line for each of dozens of special characters.
>>
>> I get these, too. I did not investigate that further yet as it seemed to
>> me that they did not have any real impact on anything. But it is on my
>> ToDO list (what isnt?)
> 
> :-) Is your ToDO list public? Maybe maintain it somewhere like
> http://techbase.kde.org/Projects/Nepomuk/DevelopmentStatus

mostly in my head. I tried to maintain something online but if no-one
else works on it it quickly gets out of date. I feel that my current
approach with continuous blog updates does work better for me.

> 
>>> 2.
>>>     [/usr/bin/nepomukservicestub] "/usr/bin/nepomukservicestub(27988)"
>>> Soprano: "Invalid argument (1)": "Cannot set values for abstract
>>> property 'http://www.semanticdesktop.org/ontologies/nfo#bitsPerSample'.
>>>
>>> I can reproduce this by running `nepomukindexer path/to/file.wav` I
>>> get it on other .wav files. What's strange is in Dolphin, the
>>> Information panel shows "Bits per sample: 24"
>>
>> Which version of shared-desktop-ontologies are you using?
> 
> Version: 0.7.0-0ubuntu1 , the stock package for Kubuntu 11.10 (Oneiric).

They should really update to 0.8.1. Nonetheless that should not be the
problem as nfo:bitsPerSample is properly defined in 0.7.

Do you have raptor and soprano's raptor parser plugin installed?

> In xmlindexer output, bitsPerSample is the one nfo property that isn't dated:
>   <value name='http://www.semanticdesktop.org/ontologies/nfo#bitsPerSample'>16
> compared with e.g.
>   <value name='http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#sampleRate'>8000
> 
> Hmmm. Both /usr/lib/strigi/strigiea_riff.so and strigita_au.so  have the string
>   http://www.semanticdesktop.org/ontologies/nfo#bitsPerSample
> and both are mmap'd by xmlindexer.
> Maybe Vishesh Handa's commit 3 months ago "Add the correct type in the
> riff analyzer The riff analyzer should add nfo:Audio or nfo:Video
> depending on the type of file it is analyzing" is relevant, but I'm
> completely guessing ;-)
> 
> 
>>> 3.
>> ...
>> This is the strigi indexer producing invalid data. This has been "fixed"
>> recently.
>>
>>>
>>> 4.
>>> "nepomukindexer dumped core indexing Firefox update file". I filed
>>> <https://bugs.launchpad.net/ubuntu/+source/kde-runtime/+bug/896456>
>>
>> This might be fixed in libstreamanalyzer master. Could you test that?
> 
> Here we go, bugs that might already be fixed. I've only got the one
> computer running Kubuntu 11.10 (KDE 4.7.2). I dunno if trying to build
> then "swap in" a few packages from master is a rat's nest. Is there a
> build service with nightly  amd64 packages?  I have a spare partition,
> I guess I could install something more bleeding-edge there and then
> build KDE from source. How do KDE developers do it?

I mix whatever I need. Of course that is not recommended if you do not
want to be solving problems all the time. :P
I will try to convince Jos to make a release.

>>> 5.
>>> I've got an Outlook message with a calendar.ics attachment that
>>> reproduceably crashes both nepomukindexer and xmlindexer in
>>> icalparser_string_line_generator(). I filed
>>> <https://bugs.launchpad.net/ubuntu/+source/kde-runtime/+bug/897529> ,
>>> the offending message is attached.
>>
>> This is interesting. This is a crash in a libstreamanalyzer plugin which
>> is part of kdepim. So a bug report against some pim component might be
>> in order - but which I am not sure...
> 
> Yup, the crash is in
> libical.so.0 from package libical0
> called from
> strigiea_ics.so  from package kdepim-strigi-plugins
> but I don't know how to reproduce from either of these. I hope the
> Kubuntu team is responsive to Launchpad bugs...
> 
> Thanks for responding!
> 
> --
> =S Page
>