Menu
▾
▴
#662 OmegaT shows incorrect match statistics for bilingual documents when source=target
OmegaT shows incorrect match statistics for segmented and pre-translated TTX files.
For No match, it shows too few words.
Most words end up in the 95%-100% category, but for 85%-94%, 5%-84%, 50%-74%, OmegaT shows no words, although there are clearly matches from these categories. See the screenshot below.
https://www.dropbox.com/s/gfh5rspn8rjajnj/incorrect%20match%20statistics%20for%20TTX.png
Will send a sample project to Didier in private.
What happens if you edit at least one segment, and save?
Is the information still wrong?
Didier
Yes, the information updates accordingly, but is wrong nonetheless.
OK. I haven't received the sample project.
Didier
The issue comes from the fact that the TTX filter add a 100% match for every segment "translated", even if the translation is English to English.
(In that specific project, every segment is "translated".)
As a workaround, one can use Rainbow to prepare an OmegaT project, which will then show correct statistics.
Didier
Fixed in SVN (/trunk).
The issue happened with all bilingual documents, including the OmegaT native PO filter. Now, when Allow translation is equal to source is not set, translations are not loaded as matches when translation is equal to source, which solves the statistics issue. The translation doesn't appear in the Comment area either.
Didier
Closing.
Fixed in the released version 3.1.1 of OmegaT.
Didier