It has been a long time ago since I first heard about strigi. It seems it has a fine integration with Nepomuk and I can enable Nepomuk and Strigi in the System Settings menu.

Both are running now and strigi indexed untill now zero files. Also the KDE search is not using any index of strigi (it is still quite slow). There are thousand possibilities for strigi and Nepomuk but it seems KDE is still not using is! Why? Both Solid and Phonon are providing now already their great capabilities, but I am not seeing anything from Akonadi and Nepomuk. For Akonadi I read Kontact is working on it (great!), but I haven't heard about an application going to use strigi (or Nepomuk).

What's the status of the development?

I'm also a bit confused about the role of Strigi / Nepomuk / Soprano. When I had a look at the speed of jpeg thumbnailing, i saw that there is code e.g. to get embedded exif thumbnails of jpegs out of the strigi analyzer. I don't know why it's not running, yet. Perhaps I misunderstood it, or some parts are simply not ready.

Yes, it's the "some parts are simply not ready"

That said, the components (strigi, nepomuk) have seen quite some nice integration in the 4.2 cycle until now. Hopefully, we can have some search functionality in the file dialogue, for example. The Nepomuk Search KIO slave is already in, btw. If you're running current trunk/, you should be able to see URLs such as nepomuksearch:// in Dolphin and Konqueror.

I agree that the possibilities, once the frameworks are firmly in place are virtually endless, but we're not quite there yet. Good things take time...

There is now a nepomuk search krunner in trunk (for KDE 4.2). So it will finaly be possible to search for files and tags (an int theory more) with ALT+F2.

There will also be a Nepomuk search KIO Slave which also means strigi. This means all KDE 4 applications get searching capabilities.

Can I just ask how much faster nepomuk is if one uses the sesame2 backend?

Cos I'm running trunk minus a week or two and it is a beast. Whenever i do any file operations copy/delete/move i.e. lots of files nepomuk just sits there for 5 minutes sucking cpu cycles on one of my cores. So is there this problem with sesame2? I'm just curious if it is worth jumping through hoops to get the sesame backend installed?

Yah, I know, I'm just lazy....

Sesame2 is much lighter, more efficient at indexing. It is not even worth having strigi enabled with the redland backend. However, since no application can currently make use of the strigi data in nepomuk, it isn't even worth having it enabled anyway, even with Sesame2, because all it's work will never be put to any use.

So I guess my massive cpu usage is from nepomuk itself rather than strigi...I guess I should just disable nepomuk for now as it's performance penalty is rather large

_txf_ wrote:So I guess my massive cpu usage is from nepomuk itself rather than strigi...I guess I should just disable nepomuk for now as it's performance penalty is rather large

Maybe you can just disable Strigi and still use Nepomuk for its other services like tagging, rating, etc.

Cheers,
_

Nepomuk tends to have a high CPU usage when strigi is indexing because strigi transfers that data into Nepomuk, which obviously gets hammered because Sesame2 does it so fast. With redland it is Strigi that abuses the CPU not nepomuk because redland is so slow.

Either way, Nepomuk without strigi runs fine, and that is how I use it.

bcooksley wrote:Nepomuk tends to have a high CPU usage when strigi is indexing because strigi transfers that data into Nepomuk, which obviously gets hammered because Sesame2 does it so fast. With redland it is Strigi that abuses the CPU not nepomuk because redland is so slow.

Oh, that is not logical and therefore not true.

Sesame2 or Redland are backends for Soprano. Soprano is what Nepomuk uses to store the data.

Soprano runs in the Nepomuk storage process. Strigi reads the data out of the files and passes the Info into Nepomuk/Soprano.

So when you use redland, Nepomuk is slow and needs a lot of CPU to save the information which stigi passes to it. And because of it being slow Strigi has to wait most of the time (which therefor does not use so much CPU and work as fast as it could).

With sesame2 Nepomuk is faster and needs less CPU per data to store it. But then it is faster so Strigi waits less time and uses more cpu.

But in both cases the Nepomuk CPU usage is high. But with Sesame2 it gets more work done in less time. (which also means with Sesame2 stirgi needs more CPU time (because it works more) as with redland).

I hope that made it clear.