Unknown reporter writes

Recoll stemming does not work well at least for Russian language. Example: File "1.txt" contains the word "кандидат" (candidate in English). File "2.txt" contains the word "кандидаты". File "3.txt" contains the word "кандидата". File "4.txt" contains the word "кандидатов".

All files are indexed OK. But word stemming during searching does not work. If I search for "кандидат" Recoll finds only "1.txt". When I search for "кандидаты" only the second file is found and so on. Stemming does not work at all, but wildcards () are OK. If I search by query "кандидат" all 4 files are found.

The "aspdict.ru.rws" file was created without errors in the ".recoll" subdirectory of my home folder. But it contains many English letter combinations. I did not find any Russian letter combinations in "aspdict.ru.rws". Of course, I do not know format of this file, but may be this information will help. Stemming is enabled for English and Russian. I tried all sensible combinations of Recoll settings related to aspell usage: they do not help.

In fact wildcards work fine and they may be used in many cases instead of stemming without problems. But I decided to inform you about the problem.


medoc writes

Thanks for this report, stemming is completely broken in 1.13, I wonder that nobody else saw it. I am going to make a new release (1.13.04) to fix this

medoc writes

Fixed by commit 1868