Incomplete/empty list of contributors in non-trunk series

[Severin H]

For a long time, the LottaNZB project only used the trunk series for translations. The upcoming 0.6 release will feature heavy changes to the translatable strings. However we'd still like to give people the ability to translate the stable 0.5 release and to make maintenance releases based on that. Therefore, I set up translations for the 0.5 series [1] and imported the existing translations.

However, I noticed that the translations in the 0.5 series only list those contributors who have made changed to the translation since the initial upload to the 0.5 series (at least that's what it looks like). Because of that, the list is incredibly incomplete. Compare [2] and [3].

This bug prevents me from exporting the translations from the 0.5 series and making a maintenance release, because most translator credits would be missing.

Is there a way to fix this?

[1] https://translations.edge.launchpad.net/lottanzb/0.5/+translations
[2] https://translations.edge.launchpad.net/lottanzb/trunk/+pots/lottanzb/fr/+details
[3] https://translations.edge.launchpad.net/lottanzb/0.5/+pots/lottanzb/fr/+details

[Severin H (severinh)]

Any thoughts on this bug? Would you recommend me to manually fix the translation credits when exporting from the non-trunk 0.5 translation series?

[Данило Шеган (danilo)]

This is an important bug, but it's not trivial to solve. What would need to happen is to copy all POFileTranslator records from existing POFiles when new 'shared' one is created, or, even better, to reconstruct them based only on contributors that are still effective on those messages in such POFiles. It might introduce performance issues if done in places which might be called from the web UI, so ideally, it would only be done on new template import (though, this would not fix the problem for existing data such as yours).

[Robert Collins (lifeless)]

This would still be good to do but we don't have further message-sharing polish in our current roadmap. We can mentor someone that wants to fix this themselves.

[Laurent Lyaudet (laurent-lyaudet)]

Hello,
This is trivial.
Steps :
- gather in a database with following tables :

create table to_translate (id, to_translate string) (parse source code bundles .mo, insert)

create table release_series (id, version_number) (parse source code bundles names, insert)

create table to_translate__release_series__link (no need to waste time on good naming, steamroller)
(id, to_translate_id FK, release_series_id FK) (parse source code bundles name .mo, insert)

create table translated (id, translated string, author nullable) (parse source code bundles .po, insert without author, cross data with your traduction software to know who is the author, beware errors and falsification, see my historic clamtk, you can ignore it first, but sooner or later you'll have also to do the data correction, instead of data cleaning, read my mind... and if you did your script correctly it is trivial and fast to rerun it on corrected data)

create table translated__release_series__link (no need to waste time on good naming, steamroller)
(id, translated_id FK, release_series_id FK) (parse source code bundles, insert)

- Enjoy...
SELECT DISTINCT translated.author -- add group by and count if you want
FROM translated
JOIN translated__release_series__link ON (translated__release_series__link.translated_id = translated.id)
WHERE translated__release_series__link.release_series_id = 111 -- or add one more join to use release_series.version_number

Best regards,
    Laurent Lyaudet

[Laurent Lyaudet (laurent-lyaudet)]

Et si vous avez une minute, vous pouvez aller un peu plus loin avec une table release et une FK sur release_series dedans...