Nepomuk search documentation - search functions and operators

[Snowhog]

'Basic' Index configuration.

Launch Recoll and click on Preferences > Index configuration
In Global parameters > Top directories click the + button to add 'top level' directories to use. By default, ~ (users /home directory) is already identified. If you don't want your email (Kontact/Kmail) to be included in the indexing, click on the + button for Skipped paths and add ~/.local/share/local-mail

[tomcloyd]

I just rebooted, and that appears to have fixed the apparent failure of recoll to attend to my explicit preferences - I'd done essentially what you suggest above. I guessed (correctly, you indicate) that "~" meant my user account, and removed that, but it clearly was indexing my email anyway, and it should not have been. I run Thunderbird, not Kmail, and that's where the email user account is kept. With a reboot, all is well.

AND, it being obvious that my test target directory tree is index, I run a query, and...wow, this is one kick-ass search tool. I CAN specify exactly what part of the filesystem to search, and the results are very nicely display, and quickly.

OK...you have a sale, sir. And my thanks. This is just peachy. So...off we go to index the whole tree. This will do what I need quite well.

[Snowhog]

You also need to set up when you want Recoll to reindex. Preferences > Indexing schedule.

[tomcloyd]

Already done. The interface is pretty clear. What wasn't clear was why it was being ignored, initially. This is hardly the first time I've seen that something doesn't work without a reboot. And I did try stopping and restarting recoll, to no effect. It just keep merrily indexing my email. Bad dog! To be fair, the indexing part of Nepomuk was also misbehaving at that time, so we clearly had some kind of underlying problem which was resolved with the reboot. I now think it had absolutely nothing to do with recoll at all. Quite likely.

I AM delighted to have this new tool to use! A good, well-focused, fast filesystem content search functionality is very very useful to me. Much more so than semantic search.

[tomcloyd]

This is by far the best indexing/search tool I've ever used. To index my whole user account typically take overnight and well into the next day (with Google Desktop or Nepomuk Desktop). Recoll did it in about 5 hours! I then asked it to find a two word phrase in a segment of the directory tree that has several hundred thousand files, at least. I had my 986 hits in about 4 seconds. The results are displayed in a very sensible way, and each is easily previewed.

The search results are displayed so quickly it's not at all clear that there's any real reason to constrain the search area except to exclude certain results. This baby is FAST.

Exactly what I need, and far more. Again, my thanks.

[Snowhog]

Glad that it is a tool that lives up to it's reputation, and that serves your needs so well.

[dmeyer]

Since you have been using it a while any chance you can give an indication of how big the database is? Such an intensive index system must surely generate quite a large database.

[Snowhog]

Here, I'm only indexing my ~ directory, currently 8.6GB out of 100GB available. This is what recoll itself is taking up:

paul@tanagra:~/.recoll/xapiandb$ ls -latotal 51056
drwxr-xr-x 3 paul paul     4096 Sep  2 18:09 .
drwx------ 3 paul paul     4096 Aug  6 21:43 ..
-rw-rw-r-- 1 paul paul        0 Sep  2 18:09 flintlock
-rw-rw-r-- 1 paul paul       28 Aug  6 21:42 iamchert
-rw-rw-r-- 1 paul paul      371 Sep  2 18:09 position.baseA
-rw-rw-r-- 1 paul paul      371 Sep  2 18:09 position.baseB
-rw-rw-r-- 1 paul paul 23101440 Sep  2 18:09 position.DB
-rw-rw-r-- 1 paul paul      335 Sep  2 18:09 postlist.baseA
-rw-rw-r-- 1 paul paul      335 Sep  2 18:09 postlist.baseB
-rw-rw-r-- 1 paul paul 20955136 Sep  2 18:09 postlist.DB
-rw-rw-r-- 1 paul paul       40 Sep  2 18:09 record.baseA
-rw-rw-r-- 1 paul paul       40 Sep  2 18:09 record.baseB
-rw-rw-r-- 1 paul paul  1581056 Sep  2 18:09 record.DB
drwxr-xr-x 2 paul paul     4096 Sep  2 18:09 stem_english
-rw-rw-r-- 1 paul paul      117 Sep  2 18:09 termlist.baseA
-rw-rw-r-- 1 paul paul      117 Sep  2 18:09 termlist.baseB
-rw-rw-r-- 1 paul paul  6594560 Sep  2 18:09 termlist.DB

paul@tanagra:~/.recoll/xapiandb$ ls -la stem_english/
total 2084
drwxr-xr-x 2 paul paul   4096 Sep  2 18:09 .
drwxr-xr-x 3 paul paul   4096 Sep  2 18:09 ..
-rw-rw-r-- 1 paul paul      0 Sep  2 18:09 flintlock
-rw-rw-r-- 1 paul paul     28 Sep  2 18:09 iamchert
-rw-rw-r-- 1 paul paul     13 Sep  2 18:09 postlist.baseA
-rw-rw-r-- 1 paul paul     31 Sep  2 18:09 postlist.baseB
-rw-rw-r-- 1 paul paul 999424 Sep  2 18:09 postlist.DB
-rw-rw-r-- 1 paul paul     13 Sep  2 18:09 record.baseA
-rw-rw-r-- 1 paul paul     25 Sep  2 18:09 record.baseB
-rw-rw-r-- 1 paul paul 630784 Sep  2 18:09 record.DB
-rw-rw-r-- 1 paul paul     13 Sep  2 18:09 termlist.baseA
-rw-rw-r-- 1 paul paul     23 Sep  2 18:09 termlist.baseB
-rw-rw-r-- 1 paul paul 466944 Sep  2 18:09 termlist.DB
paul@tanagra:~/.recoll/xapiandb$

[dmeyer]

That's actually not bad at all. I would be interested if tomcloyd who I assume to have a big home file has a much larger database. I am definitely going to give recoll a go over the weekend but I have an SSD and I try my best to keep it safe.

Thanks

[vinnywright]

my ~/.recoll/xapiandb DER is 351.9 MiB indexing a 195Gig ~/ of witch 145Gig's is used ,gives me instant results to a search with decent preview options not bad I would say

VINNY

[vinnywright]

just out of curiosity I just did a search for "the hobit" in both recoll and dolphin ,,recoll instantly gave me a " Alternate spellings (accents suppressed): hobbit " ,,,, dolphin took about 30seconds to decide to display nothing .

redoing the search in both applications with "the hobbit" returned a instant result in recoll and though dolphin did find the file and directory it again took around 30 seconds or more to display anything.

VINNY

[dmeyer]

Dolphin isn't a very good searcher, even with the new improvements, that's why I always recommend nepoogle. Still recoll seems helluva impressive. I'm going to install recoll this weekend and get a grasp of it. It has a Krunner too :-D

By the way, I still think people miss the point. Nepomuk isn't about content but about context. Its difficult to grasp but when a person does, you realize its brilliant at what it does but it sucks at traditional search. I'm wondering if there isn't a way to somehow integrate recoll into dolphin. That would be epic. I see it has some konqueror integration but I find konqueror antiquated.