#9315 closed defect (fixed)
Used old internationalization file for Japanese language.
Reported by: | mfuji | Owned by: | team |
---|---|---|---|
Priority: | normal | Milestone: | 13.11 (hotfix) |
Component: | Core | Version: | tested |
Keywords: | i18n | Cc: | simon04 |
Description
Jar file of version 6383 seems to contain an old i18n file for Japanese.
It contains many wrong translation which were already fixed months ago.
Attachments (0)
Change History (27)
comment:1 Changed 10 years ago by
Cc: | simon04 added |
---|
comment:2 Changed 10 years ago by
comment:3 Changed 10 years ago by
Resolution: | → fixed |
---|---|
Status: | new → closed |
Actually, I have no idea what happened during the i18n process. It seems that old .po
files have been retrieved from launchpad since ja.po
contains:
"POT-Creation-Date: 2013-11-12 10:58+0100\n" "PO-Revision-Date: 2012-05-27 13:59+0000\n" ... "X-Launchpad-Export-Date: 2012-06-02 04:48+0000\n" "X-Generator: Launchpad (build 15342)\n"
The one used for r6387 seems to be correct:
"POT-Creation-Date: 2013-11-15 21:07+0100\n" "PO-Revision-Date: 2013-11-12 08:22+0000\n" ... "X-Launchpad-Export-Date: 2013-11-15 04:48+0000\n" "X-Generator: Launchpad (build 16831)\n"
comment:4 Changed 10 years ago by
Resolution: | fixed |
---|---|
Status: | closed → reopened |
This still does not seem finished. Before this patch, the 'sk' translations was shown to be 76% complete at http://josm.openstreetmap.de/wiki/Translations (and also JOSM start page). Now it is at 89%. But that is still bogus as the 'sk' translation should be at 99.9% (JOSM core, not extensions). Please check it.
comment:5 follow-up: 8 Changed 10 years ago by
At least the used sk.po
seems okay:
"POT-Creation-Date: 2013-11-15 21:07+0100\n" "PO-Revision-Date: 2013-11-12 19:12+0000\n"
I'm unsure whether the statement "JOSM core, not extensions" on Translations is correct since the entire translation template (w/ plugins) contains 8397 msgid
s, among those 2378 belong to plugins.
Do you recall the percentage stated on Translations before r6383? Besides the percentage, are there other indicators that the sk
translations are broken/incomplete/outdated?
comment:6 Changed 10 years ago by
Was Japanese the only language impacted ? If not we should release a new tested :)
comment:7 Changed 10 years ago by
Apparently not (there are (too) many files from 2012):
josm.all/i18n/po$ grep PO-Revision-Date *.po bg.po:"PO-Revision-Date: 2013-06-24 15:10+0000\n" ca.po:"PO-Revision-Date: 2013-11-11 07:36+0000\n" cs.po:"PO-Revision-Date: 2013-10-21 06:58+0000\n" da.po:"PO-Revision-Date: 2012-06-01 08:35+0000\n" de.po:"PO-Revision-Date: 2013-11-08 06:49+0000\n" el.po:"PO-Revision-Date: 2013-08-24 07:36+0000\n" en_AU.po:"PO-Revision-Date: 2011-07-20 11:31+0000\n" en_GB.po:"PO-Revision-Date: 2012-04-11 21:53+0000\n" es.po:"PO-Revision-Date: 2012-05-01 11:44+0000\n" et.po:"PO-Revision-Date: 2012-05-23 20:38+0000\n" eu.po:"PO-Revision-Date: 2013-03-30 13:56+0000\n" fi.po:"PO-Revision-Date: 2012-04-10 16:28+0000\n" fr.po:"PO-Revision-Date: 2013-10-21 16:21+0000\n" gl.po:"PO-Revision-Date: 2013-03-30 13:37+0000\n" hu.po:"PO-Revision-Date: 2013-11-10 00:31+0000\n" id.po:"PO-Revision-Date: 2011-12-14 05:58+0000\n" it.po:"PO-Revision-Date: 2013-11-11 07:12+0000\n" ja.po:"PO-Revision-Date: 2012-05-27 13:59+0000\n" nb.po:"PO-Revision-Date: 2011-07-19 17:53+0000\n" nl.po:"PO-Revision-Date: 2012-03-04 12:02+0000\n" pl.po:"PO-Revision-Date: 2013-11-01 10:58+0000\n" pt_BR.po:"PO-Revision-Date: 2013-10-11 12:18+0000\n" pt.po:"PO-Revision-Date: 2012-06-02 00:32+0000\n" ru.po:"PO-Revision-Date: 2012-06-01 13:53+0000\n" sk.po:"PO-Revision-Date: 2012-05-15 09:12+0000\n" sv.po:"PO-Revision-Date: 2012-05-20 09:59+0000\n" tr.po:"PO-Revision-Date: 2013-11-05 19:02+0000\n" uk.po:"PO-Revision-Date: 2013-11-10 16:26+0000\n" zh_CN.po:"PO-Revision-Date: 2012-04-16 12:46+0000\n" zh_TW.po:"PO-Revision-Date: 2013-07-14 15:08+0000\n"
comment:8 follow-up: 9 Changed 10 years ago by
Replying to simon04:
At least the used
sk.po
seems okay:
"POT-Creation-Date: 2013-11-15 21:07+0100\n" "PO-Revision-Date: 2013-11-12 19:12+0000\n"
Yes, I have now downloaded both 6386 and 6388 and checked that 6386 contains old outdated sk translation. 6388 seems to contain the current state of the translation so it is fine.
I'm unsure whether the statement "JOSM core, not extensions" on Translations is correct since the entire translation template (w/ plugins) contains 8397
msgid
s, among those 2378 belong to plugins.
So this claim on the Translation page should be sorted out.
Do you recall the percentage stated on Translations before r6383? Besides the percentage, are there other indicators that the
sk
translations are broken/incomplete/outdated?
I recall that the percentage is going down with each tested release which I find strange because I translate more and more strings (in core). But maybe it is just that the plugins are inflating faster than I can improve core... I do not see any other indicator that the translation is broken.
I also think you should upload a new 'tested' version (but based on 6383+fixed translations, not 6388 with already changes English strings).
comment:9 Changed 10 years ago by
Replying to aceman:
Yes, I have now downloaded both 6386 and 6388 and checked that 6386 contains old outdated sk translation. 6388 seems to contain the current state of the translation so it is fine.
Good to hear that :-)
So this claim on the Translation page should be sorted out.
I agree.
I recall that the percentage is going down with each tested release which I find strange because I translate more and more strings (in core). But maybe it is just that the plugins are inflating faster than I can improve core... I do not see any other indicator that the translation is broken.
This might be due to many plugins being added. In #8645 we are discussing to split the translation into core, plugins and maybe data (for presets, imagery). Feel free to join the discussion.
I also think you should upload a new 'tested' version (but based on 6383+fixed translations, not 6388 with already changes English strings).
So r6387 should be released as new tested
…
comment:11 Changed 10 years ago by
I am unsure what led to this error. In past times, I never had troubles with those updates. It remains to decide on the transifex topic. ;-)
comment:12 Changed 10 years ago by
comment:13 follow-up: 14 Changed 10 years ago by
Resolution: | → fixed |
---|---|
Status: | reopened → closed |
I wasn't aware we couldn't promote intermediate versions (those between latest) as tested, so it didn't work. I have changed it to r6388 and launched the build manually, it worked.
comment:14 Changed 10 years ago by
Replying to Don-vip:
I wasn't aware we couldn't promote intermediate versions (those between latest) as tested, so it didn't work. I have changed it to r6388 and launched the build manually, it worked.
Yep. That's not possible. A tested must always have been a latest before.
I'm unsure whether the statement "JOSM core, not extensions" on Translations is correct since the entire translation template (w/ plugins) contains 8397 msgids, among those 2378 belong to plugins.
So this claim on the Translation page should be sorted out.
If these texts are wrong, then the translation data files are wrong as well. These numbers are directly taken from the included data files.
comment:15 Changed 10 years ago by
No, the data files are fine and the numbers are correct too. Opening the .po file for 'sk' in poedit also claims to have 89% translated. But that those are all strings including plugins.
The problem is the wiki page above that table claims the numbers are without plugins. So if you not do any processing of the data files to remove plugins (and count the percentage anew), then the claim is wrong.
comment:16 Changed 10 years ago by
The data files contains these many strings. When I recreate them they only contains about 6200.
Something went wrong with creation of the language files. They include too many strings.
comment:17 follow-up: 18 Changed 10 years ago by
Last known good .po file I have stored locally has 8349 English strings (according to poedit program). Not sure how it counts plural forms.
comment:18 Changed 10 years ago by
Replying to aceman:
Last known good .po file I have stored locally has 8349 English strings (according to poedit program). Not sure how it counts plural forms.
This is not about .po files. These are translated into ".lang" files. The Wiki page shows the contents of these lang files, which are directly used by JOSM.
comment:19 follow-up: 20 Changed 10 years ago by
Sure, but Launchpad should export/import all those strings into the .po file 1:1 and not miss any strings in the process. So I hope the .lang and .po files are equivalent in content. If not, there is a problem.
comment:20 Changed 10 years ago by
Replying to aceman:
Sure, but Launchpad should export/import all those strings into the .po file 1:1 and not miss any strings in the process. So I hope the .lang and .po files are equivalent in content. If not, there is a problem.
They are not. The .lang files only contain the part relevant for the software (i.e. core or plugin). Thus it should always be a subset of the whole .po file.
comment:21 follow-up: 22 Changed 10 years ago by
I could only find one .lang file per language in the data/ subdirectory of JOSM source.
But anyway, what is the conclusion. Do you see any problem?
comment:22 follow-up: 24 Changed 10 years ago by
Replying to aceman:
But anyway, what is the conclusion. Do you see any problem?
As Vincent did the update I think he needs to verify (next time :-) that his build environment is ok. There are some cases which cause a bit trouble for the i18n stuff (e.g. adding additional files into the source tree or adding links, ...). Maybe we also could automate the i18n update on the server to eleminate such troubles.
As long as people or computers are involved, errors will happen :-)
comment:24 Changed 10 years ago by
Replying to stoecker:
As Vincent did the update I think he needs to verify (next time :-) that his build environment is ok.
Hey I didn't do the i18n update this time ! :D (but I already broke it once in the past, it seems easy to make it bad)
Maybe we also could automate the i18n update on the server to eleminate such troubles.
+1 . If we could automate the entire thing with verification checks it would be a huge benefit :)
comment:25 Changed 10 years ago by
I performed the i18n update as summarized in ticket:8645#comment:17. Not sure why this let to problems.
comment:26 Changed 10 years ago by
Milestone: | → 13.11 (6388 hotfix) |
---|
In 6387/josm: