Issue 110: python api - no stemming?

Heya,

I get the exact same results for

----query.execute(qs, stemming=0) ---- as ----query.execute(qs, stemming=1) ---- whereas with the gui or ----recoll -t <qs > ---- I get plenty more, obviously due to inclusion of different variations.

Any hints?

It seems that the stemming language is hard-coded at english in the Python module (yes, silly). I’m going to fix this, but, just to confirm that this is the issue, what stemming language are you using ?

English.

Also - not sure whether it matters - but the recoll version I’m on is ----1.17.3-2 ---- (debian package), with not much hope to change that unless I get a toolchain for raspberrypi together. I will however build 1.18 on x86 to check if it’s the same.

Riiight, with ----recoll 1.18 ---- the results from python are the same as results from ----recoll -t ---- and do include stemmed words - success! I guess this has been fixed then. The bad news is that 1.18 is only available in the ----experimental ---- Debian repo which is unavailable in raspbian, I guess that toolchain will be coming along then..

Yes, you are right, stemming was always off in 1.17, not hardcoded to english, as it is in 1.18.1. It means that stemming is off in the Ubuntu lens, and nobody ever complained…

This is really sloppy, my apologies. I think that recoll 1.18 is currently in experimental because of the Wheezy freeze. I guess that the Wheezy release could still take some time and that there is no chance to get a fix into it, so your best bet would probably be to find a way to cross-compile for the raspberry, rather than wait for a future raspbian release.

Also, the patch to allow specifying the stemming language is a simple one, I’ll apply it to the future 1.18 maintenance release. Closing this issue as there is nothing more I can do about it (fixing 1.17 would be easy, but I don’t see that it could make any good at this point).

Fixed in the latest code.