[SOLVED] Nepomuk File Indexing Service Not Running

[T.C. Hollingsworth]

On Mon, Dec 26, 2011 at 11:06 AM, Marko Vojinovic <vvmarko at gmail.com> wrote:
> On Monday 26 December 2011 17:39:12 Anne Wilson wrote:
>> On 12/26/2011 10:41 AM, Marko Vojinovic wrote:
>> > Nepomuk and Strigi are one of the first things I shut down when I
>> > install a fresh Fedora with KDE. On every login they would start
>> > indexing the "desktop" or whatever (I have nothing bar two default
>> > icons in my Desktop folder), and then the virtuoso binary would start
>> > eating my CPU time for half-a-minute, before finally settling down. And
>> > I see absolutely no (obvious) use of all that on my machine.
>>
>> Nepomuk *should* not be a problem - strigi is the indexer, and is where
>> the resource hog could be expected.  AIUI, Strigi indexes the
>> directories that you give access to, and Nepomuk allows all
>> Ankonadi-aware applications to draw on the information in the databases.
>>  Of course this is not a complete description, but until someone comes
>> up with a better one....
>
> Right, but it still doesn't make much sense to me. If strigi is shut down,
> there is nothing in the database, so what would be the purpose of nepomuk to
> keep running in that case?

Nepomuk does more than file search.  You could have info about tagged
photos, artists in music libraries, etc. stored in it.

> Also, is strigi indexing the contents of files, or just filenames?

Yes.

> Can nepomuk
> be configured to use the mlocate database instead (or in addition to strigi)?

Not that I'm aware of.  IIUC mlocate deals solely with filenames,
while strigi operates on all sorts of data, so I'm not sure there's
much benefit to doing so.

> If strigi indexes the file contents, what types of files can it read through?
> Can it index anything other than ASCII files, like libreoffice, pdf, djvu, tags
> in mp3, metadata in various audio/video files?

Yes.

> Can it be used in conjunction
> with some OCR software to index scanned handwriting in jpeg files?

I'm not aware of any OCR software that handles handwriting very well,
nor would I want it to start randomly OCRing every JPEG I own.  I
don't think there's anything stopping someone from writing an app that
OCRs documents and populating Nepomuk with the text, if someone finds
that useful.

> Is it
> multilingual, ie. does it index stuff written in Cyrillic/Chinese/Hebrew/etc?

Of course!  KDE is generally very good about that kind of thing.

> Next, which apps are "akonadi-aware"? How could they use the data indexed by
> strigi, and to what purpose?

All of the KDEPIM apps:  KMail, Kontact, etc.  For searching and fast
filtering, I presume

> I'd really like to see a real-world example of all this being useful.
> Otherwise it appears like a solution looking for a problem.

Part of the problem is that it has some dumb defaults.  For instance,
it only searches the Documents, Music, Movies, etc. folders that I
rarely use.  I always configure it to search my entire home directory.
 Another thing that annoys me is that KRunner doesn't do desktop
search by default.  Once that's enabled, KRunner becomes *a lot* more
useful (e.g. more than just slightly faster than launching something
on a terminal ;-).

Instead of manually browsing to a directory and opening a particular
file, I just press ALT+F2, enter the first few characters, and hit
Enter.  I can search for artists in my music, pictures of my
boyfriend, functions in source code, and start programs all in one
text box.

> I can use locate to search through file names, and grep to search through file
> contents (even through binary files :-) ). So I expect that
> strigi/nepomuk/akonadi-aware app should be able to do more than that, right?

For the same reason Google searches an index they build rather than
"grepping" the Internet on the fly (or even their cached copy,
presumably).  It's more performant when you're searching a large
amount of data.  I don't know about yours, but "grep 'something' ~"
would take hours on my machine.  ;-)

If I'm looking for a particular function in a codebase, I'll probably
still use grep.  But, if I'm looking for some random letter I wrote
six months ago and not sure where I stuck, I'd use KDE's desktop
search.

-T.C.