Page MenuHomePhabricator

Tux does not always show 100% matches that should be in translation memory
Closed, DuplicatePublicBUG REPORT

Description

Steps to replicate the issue (include links if applicable):

From a *backup copy* of Meta-Wiki around but before 2022-11-26:

I’m sorry I can’t describe any general way to replicate the issue. The issue is intermittent and I’m not seeing a pattern so I really don’t know how to replicate it. If you need live data I’ll check to see if the issue affects next week’s Tech News; but we only have 3 to 4 days to work on Tech News so if the issue cannot be replicated within that window I’ll have to proceed with the translation anyway.

What happens?:
Note that “suggestions” for “Recurrent item” and “Future changes” do not show the most up-to-date translations (i.e., the translation used in week 47 or 46). Sometimes suggestions will come up completely blank even though the exact same string had been translated the week before.

What should have happened instead?:
For those cases where a match should have been found but not, matches should have been “suggested”.

For those cases where outdated “suggestions” are shown, suggestions should not be weighted only by number of times used but also by date; recent translations should be given more weight. As it is, we can’t tell if any particular suggestion is old or new so we have to ignore all suggestions.

Implication: Because we can’t trust the suggestions list, we are forced to find a suitable translation by manually looking for one in a previous week, go to the translation screen for that page, then copy translations by hand. This completely defeats the whole purpose of the “suggestions” feature and in fact defeats the whole point of having a TM at all.

Software version (skip for WMF-hosted wikis like Wikipedia):

Other information (browser name/version, screenshots, etc.):
This might be specific to Meta-Wiki. I don’t have the background knowledge to tell if this is a general issue or an issue specific to Meta.

Event Timeline

Tacsipacsi renamed this task from CAT does not always show 100% matches that should be in translation memory to Tux does not always show 100% matches that should be in translation memory.Nov 27 2022, 6:40 PM
Tacsipacsi subscribed.

By the way, I’m doing this week’s Tech News right now and there are like at least 3–4 strings (which I’ve been using for a few weeks, basically since I’ve been doing this) that have no valid suggestions at all (i.e., either no suggestions or only 1 suggestion from way back before I started). Something is very wrong with the matching algorithm.

This is continuing to happen this week. Some of these failures are pretty inexplicable. For example, my translation for “Recent changes” is still not showing as a “suggestion” even though it’s recent AND has been used for at least half a dozen times already. Meanwhile, ancient translations that have been been used for only 2 times are showing.

Example:

This is last week’s Tech News:

Screenshot_2022-12-10_19-04-27.png (1×886 px, 183 KB)

This is this week’s. As you can see, the translation used last week isn’t showing. I had to copy the correct translation by hand to the text box.

Screenshot_2022-12-10_19-04-49.png (720×939 px, 110 KB)

Neither is “Changes later this week” although my translation has by now been used at least a few times. Meanwhile, two old translations that I had rejected are shown, even though they have been used only once.

Screenshot_2022-12-10_19-05-10.png (1×941 px, 134 KB)

“Recurrent item” exhibits the same problem: Old, rejected strings are shown; the new, currently used string isn’t in the list at all.

Screenshot_2022-12-10_19-05-28.png (1×942 px, 131 KB)