Bug 1138644 - Translated strings became untranslated after the Anaconda update is pushed to Transifex
Summary: Translated strings became untranslated after the Anaconda update is pushed to...
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Samantha N. Bueno
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2014-09-05 11:09 UTC by TianShixiong
Modified: 2014-10-11 10:56 UTC (History)
8 users (show)

Fixed In Version: anaconda-22.9-1
Clone Of:
Environment:
Last Closed: 2014-10-10 20:18:03 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
Anaconda untranslated strings(previously translated) (15.82 KB, image/png)
2014-10-09 14:32 UTC, TianShixiong
no flags Details
Untranslated string(previously translated) (24.66 KB, image/png)
2014-10-09 14:33 UTC, TianShixiong
no flags Details

Description TianShixiong 2014-09-05 11:09:20 UTC
Description of problem:
The recent two or three updates of Anaconda(Master) didn't make any changes in some strings, I compared the old strings with new updated one, they are 100% the same. However the translations of these strings are left to blanks(not translated) after they were pushed to Transifex. 

For example:

/boot filesystem

The installation was stopped due to incomplete spokes detected while running in non-interactive cmdline mode. Since there can not be any questions in cmdline mode, edit your kickstart file and retry installation. The exact error message is: %s. The installer will now terminate.

bootloader stage2 device %(stage2dev)s is on a multi-disk array, but bootloader stage1 device %(stage1dev)s is not. A drive failure in %(stage2dev)s could render the system unbootable. 

in file: pyanaconda/bootloader.py:1362 

The following error occurred while installing the bootloader. The system will not be bootable. Would you like to ignore this and continue with installation? 
Size:

in file: pyanaconda/errors.py:245 

and so on. 

Totally 77 unchanged strings this time. It's about 80 strings last time.

I remember that these strings are the same strings that lost their translations in last update. 



Version-Release number of selected component (if applicable):
I think it should be anaconda-21.48.4-1 according to the submit time.

How reproducible:


Actual results:
The translations of unchanged strings after an update were lost, they became to 'untranslated' strings.

Expected results:
The strings that are not changed during the process of Anaconda development, their translations should be kept after the update is pushed to Transifex(or Zanata in the future).

Additional info:
We translators have to use Transifex feature "use this" button(even copy and paste) to restore these translations one by one. It's really wasting of time. 
It't very lucky that Transifex keeps these history translations, if not, then all of our hard work will lost.

So you guys please take some time to find out the reason and fix it.

I'd better stop translating these unstranslated strings in this weekend, and wait for your answers or suggestions.

Comment 1 Samantha N. Bueno 2014-09-05 14:12:33 UTC
I'm very sorry about that. I accidentally skipped a critical step when I branched anaconda for f21. This should not be a problem anymore.

Comment 2 TianShixiong 2014-09-05 15:22:13 UTC
(In reply to Samantha N. Bueno from comment #1)
> I'm very sorry about that. I accidentally skipped a critical step when I
> branched anaconda for f21. This should not be a problem anymore.

Thanks for the quick response.
Can you restore the master branch to last 100% translated status? Cause I didn't find any new changes there. Untranslated 77 strings are not changed comapred its last update.

Comment 3 TianShixiong 2014-09-06 04:24:51 UTC
(In reply to TianShixiong from comment #2)
> (In reply to Samantha N. Bueno from comment #1)
> > I'm very sorry about that. I accidentally skipped a critical step when I
> > branched anaconda for f21. This should not be a problem anymore.
> 
> Thanks for the quick response.
> Can you restore the master branch to last 100% translated status? Cause I
> didn't find any new changes there. Untranslated 77 strings are not changed
> comapred its last update.

Don't restore them again. I manually restored them, and fixed some translations too.
They are 100% now.

Comment 4 TianShixiong 2014-09-11 11:24:31 UTC
Hi guys, the same problem happened again today.

Please, fix it again and don't let it happen again please.

I am not sure about the version. Maybe its 22.5-1, cause I see an update log here[1]. 

[1] https://git.fedorahosted.org/git/anaconda.git

Comment 5 TianShixiong 2014-10-07 12:33:06 UTC
Any update?
The problem is still there.

Comment 6 Piotr Drąg 2014-10-08 15:25:24 UTC
I think the same problem affects pykickstart and blivet.

Comment 7 Brian Lane 2014-10-08 18:48:48 UTC
(In reply to Piotr Drąg from comment #6)
> I think the same problem affects pykickstart and blivet.

If so then please open bugs with details. This has been elusive. I think it is fixed for f21. For example, I looked at the history of several languages on the f21-branch yesterday, focusing on incomplete translations and none of them looked like they were clobbered by last weeks build.

If you still see a problem please describe it in precise detail, including how you view whatever it is that is wrong.

Comment 8 Samantha N. Bueno 2014-10-08 19:34:08 UTC
I agree with Brian. I'm really baffled on this, to be hones. In anaconda's f21-branch, for instance, .tx/config is:

$ cat .tx/config 
[anaconda.f21-branch]
file_filter = po/<lang>.po
source_file = po/anaconda.pot
source_lang = en
type = PO

[main]
host = https://www.transifex.com

Whereas on master branch (for rawhide), .tx/config is:
$ cat .tx/config 
[anaconda.master]
file_filter = po/<lang>.po
source_file = po/anaconda.pot
source_lang = en

[main]
host = https://www.transifex.com

So these are configured properly.

Comment 9 TianShixiong 2014-10-09 14:32:18 UTC
Created attachment 945330 [details]
Anaconda untranslated strings(previously translated)

Problem still there.
The screenshot is one of the string which was translated before, but is marked to untranslated string after push update. And you can see that there is a 100% same suggestion there.

I am not sure whether these screenshots can explain the precise details. If you need more info, please tell me how to collect them, and I will do it.

Comment 10 TianShixiong 2014-10-09 14:33:10 UTC
Created attachment 945331 [details]
Untranslated string(previously translated)

Comment 11 Brian Lane 2014-10-09 16:51:15 UTC
Are you sure you are looking at the right thing? The way I'm checking this is:

1. go to the f21-branch of Anaconda in transifex
2. Select zh_CN (this is an assumption, I'm not sure exactly which one you are looking at)
3. Search for fstab in the text
4. I see a translation for both places where fstab are used, including the one in your screenshot.
5. Looking at history I see a change from sbueno from a month ago.

Everything looks fine to me. How does your process for confirming it differ from mine?

Comment 12 TianShixiong 2014-10-10 15:33:05 UTC
(In reply to bcl from comment #11)
> Are you sure you are looking at the right thing? The way I'm checking this
> is:
> 
> 1. go to the f21-branch of Anaconda in transifex
> 2. Select zh_CN (this is an assumption, I'm not sure exactly which one you
> are looking at)
> 3. Search for fstab in the text
> 4. I see a translation for both places where fstab are used, including the
> one in your screenshot.
> 5. Looking at history I see a change from sbueno from a month ago.
> 
> Everything looks fine to me. How does your process for confirming it differ
> from mine?

I am sorry that I didn't mention clearly, the screenshots I made are from master branch.

In f21-branch, I followed your steps, and everything looks the same with yours. For example, the string "There is an entry in your /etc/fstab file that contains an invalid or incorrect filesystem type:" is translated.

However, it's not translated in master branch. No history record. But it had been 100% well translated about 4 times before I report this bug. This the point of the problem. Several strings(about 80 I think) like this one are repeatedly marked to untranslated after each push to master branch.

Comment 13 Brian Lane 2014-10-10 16:49:30 UTC
Ah! This makes sense now then -- when f21 had not been branched yet it broke some of the master translations.

Sadly, I don't think there is a solution other than re-doing those translations. We can't just grab the old anaconda.pot file since master has moved on from f21 and added other strings. I think the only solution is to redo the translations.

I'm going to look at adding a check to the build process that makes sure the transifex branch matches the build branch before pushing a new anaconda.pot is pushed.

Comment 14 Brian Lane 2014-10-10 20:18:03 UTC
I have added a check to our build scripts that will stop it if the transifex branch doesn't match the git branch. This should prevent this problem from happening in the future. I'm marking it as closed and apologize for any extra work it has imposed on translators. Keep up the good work.

Comment 15 Piotr Drąg 2014-10-11 10:56:46 UTC
(In reply to bcl from comment #7)
> (In reply to Piotr Drąg from comment #6)
> > I think the same problem affects pykickstart and blivet.
> 
> If so then please open bugs with details.

Fwiw, filed as bugs #1151749 and #1151750.


Note You need to log in before you can comment on or make changes to this bug.