Localization files not being used in Minetest releases

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Localization files not being used in Minetest releases

by Hamlet » Wed Jun 27, 2018 13:06

Minetest v0.4.17.1
.../minetest/po/it/minetest.po at line 10:
"PO-Revision-Date: 2017-06-03 10:10+0000\n"
(77% translated)

Now, if you go on https://hosted.weblate.org/projects/min ... netest/it/ and check minetest-minetest-it.po at line 6:
"PO-Revision-Date: 2017-09-29 19:47+0000\n"
(100% translated)

As we all know Minetest v0.4.17 has been released on June 2018.

Question #1: is localization no longer supported?
Question #2: if localization is still supported, should translators use PRs on Git instead of Weblate?
Question #3: if localization is still supported, will it be released a Minetest v0.4.17.2 with the up to date localization files?
Question #4: if localization is still supported, how to contribute translations for Minetest v5 before that it get released?
 

User avatar
rubenwardy
Moderator
 
Posts: 5890
Joined: Tue Jun 12, 2012 18:11
Location: United Kingdom
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy

Re: Localization files not being used in Minetest releases

by rubenwardy » Wed Jun 27, 2018 13:37

0.4.17 is a bug fixing release, and is a backport. Weblate is for 5.0 not 0.4.17, so it doesn't make sense to merge it into 0.4.17.

1) of course it is, 0.4.17 just doesn't include translation updates
2) No.
3) No.
4) On weblate
 

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Re: Localization files not being used in Minetest releases

by Hamlet » Wed Jun 27, 2018 16:07

rubenwardy wrote:0.4.17 is a bug fixing release, and is a backport.


v0.4.17.x is the current latest release being publicized on the Minetest's website front page, 5-stable does not exist.

rubenwardy wrote: Weblate is for 5.0 not 0.4.17,


This is false: the .po files on Weblate are the ones used for 0.4.17, else they would have strings like "Content" and "Browse online content" used in 5-dev and Weblate would mark the file itself as partially translated instead of 100% translated.

EDIT (screenshot):
Image

rubenwardy wrote: so it doesn't make sense to merge it into 0.4.17.


Your previous statement is false, so, again, v0.4.17.x is the current latest release being publicized on the Minetest's website front page, 5-stable does not exist thus it would make sense to update its locale files given that they are obsolete.

rubenwardy wrote:1) of course it is, 0.4.17 just doesn't include translation updates


Given that from 0.4.16 to 0.4.17 has passed one year I would be ashamed of ignoring the work that translators have done in the meanwhile.

rubenwardy wrote:2) No.
3) No.
4) On weblate


Thank you for answering to my questions.
If this is the official developers' position and policy - and I do believe so - toward translators and translations then I did well deleting my Weblate account.
Don't waste your time replying to this post, as I won't waste my time anymore doing useless translations.
Attachments
weblate_screenshot.png
(39.48 KiB) Not downloaded yet
 

User avatar
Linuxdirk
Member
 
Posts: 2226
Joined: Wed Sep 17, 2014 11:21
Location: Germany
In-game: Linuxdirk

Re: Localization files not being used in Minetest releases

by Linuxdirk » Wed Jun 27, 2018 16:34

Hamlet wrote:Question #2: if localization is still supported, should translators use PRs on Git instead of Weblate?

Weblate is the default for collaborative translation work. There is nothing wrong with it. No-one in the right mind uses anything else for that specific task.
 

User avatar
Krock
Developer
 
Posts: 4412
Joined: Thu Oct 03, 2013 07:48
Location: Switzerland
GitHub: SmallJoker

Re: Localization files not being used in Minetest releases

by Krock » Wed Jun 27, 2018 16:40

Hamlet wrote:This is false: the .po files on Weblate are the ones used for 0.4.17, else they would have strings like "Content" and "Browse online content" used in 5-dev [..]

This is true: The 0.4.16 translations (.po files) are used for 0.4.17(.1) because the development used to be on the master branch.
This is false: The translations for 5.0.0 are simply not updated yet. Expect more incomplete translations as soon it's done.

Hamlet wrote:Your previous statement is false, so, again, v0.4.17.x is the current latest release being publicized on the Minetest's website front page, 5-stable does not exist thus it would make sense to update its locale files given that they are obsolete.

Your statement is false, so again, 5.0.0 is developed on the master branch and 0.4.17(.1) is only a backport, which happened on a separate branch, which cannot be translated by Weblate (limited to master). I believe it's acceptable when a few strings aren't translated. You can update your locale manually using the Weblate git repository if you'd like to do that.

Hamlet wrote:Given that from 0.4.16 to 0.4.17 has passed one year I would be ashamed of ignoring the work that translators have done in the meanwhile.

If there were changes to 0.4.17(.1) then it's a pity - too bad wasn't done when 0.4.16 came out. However, their work is far from being lost. It'll be included in 5.0.0 as soon it's out.

Hamlet wrote:If this is the official developers' position and policy - and I do believe so - toward translators and translations then I did well deleting my Weblate account.

Too bad to see you go. Thanks for the translation contributions you did.

Hamlet wrote:Don't waste your time replying to this post, as I won't waste my time anymore doing useless translations.

Oops.
Look, I programmed a bug for you. >> Mod Search Engine << - Mods by Krock - DuckDuckGo mod search bang: !mtmod <keyword here>
 

User avatar
Linuxdirk
Member
 
Posts: 2226
Joined: Wed Sep 17, 2014 11:21
Location: Germany
In-game: Linuxdirk

Re: Localization files not being used in Minetest releases

by Linuxdirk » Wed Jun 27, 2018 17:15

Krock wrote:[Translations] be included in 5.0.0 as soon it's out.

Does that apply to default and Minetest Game’s mods, too?
 

User avatar
Krock
Developer
 
Posts: 4412
Joined: Thu Oct 03, 2013 07:48
Location: Switzerland
GitHub: SmallJoker

Re: Localization files not being used in Minetest releases

by Krock » Wed Jun 27, 2018 18:36

Linuxdirk wrote:Does that apply to default and Minetest Game’s mods, too?

I don't know when (or if) translations will come for MTG. Client-sided translations will be supported but the helper scripts PR needs to be added and the fallback translations need fixing. However, a PR to begin the translations might give other people the motivation to contribute to it.
Look, I programmed a bug for you. >> Mod Search Engine << - Mods by Krock - DuckDuckGo mod search bang: !mtmod <keyword here>
 

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Re: Localization files not being used in Minetest releases

by Hamlet » Thu Jun 28, 2018 12:52

Linuxdirk wrote:
Hamlet wrote:Question #2: if localization is still supported, should translators use PRs on Git instead of Weblate?

Weblate is the default for collaborative translation work. There is nothing wrong with it. No-one in the right mind uses anything else for that specific task.


If they used it instead of letting it rot you would be right. Since they don't seem to look at anything else than PRs then maybe they wouldn't miss translation updates that way.
 

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Re: Localization files not being used in Minetest releases

by Hamlet » Thu Jun 28, 2018 12:54

My point of view:

Dev1 "Are we going to release 0.4.17?"
Dev2 "Yes."
Dev1 "Any translations have been updated?"
Dev2 "Yes."
Dev1 "Then let's merge them and release 0.4.17"
Dev2 "Merged, let's release."

Instead:

Dev1 "v0.4.17 has been released."
User "You haven't updated the translations."
Dev1 "Uh, well, that's because those files are for v5."
User "Actually they are still for 0.4.16, one year old."
Dev1 "Uh, oh, no you are wrong, those are for v5, just not updated."
User "So since 0.4 is the current stable version are you going to merge those translations?"
Dev1 "Of course not, it would not make any sense since they are for v5, and anyway it would be impossible because of very complex technical reasons."
User "Actually they are still for 0.4.16, one year old."
Dev1 "You are wrong, believe me. Furthermore 0.4.17.2 is taboo."
User "So you are not going to merge those v5-notupdated-translations into 0.4?"
Dev1 "No, absolutely impossible, they are for v5, we'll merge them in v5."
Dev1 "Bugfix release: no new features."
User "But translations are not 'features'."
Dev1 "You are wrong, they are features."
User "... out of curiosity, 1+1 = ?"
Dev1 "3, of course."
User "Wasn't it 2?"
Dev1 "No, you're wrong."

:DDD
Sorry, I don't buy that.
Weblate's .po files are v0.4, current stable version is 0.4; 1+1=2
Please don't tell me that 1+1=3, I am not going to ever buy it.

It would be honest to simply admit that translations have been completely forgotten and now for some reason you don't want to take the hassle of releasing a 0.4 version shipping the updated translations.

You can't think that people doesn't feel like being fooled after that they've been waiting for one year just to see their translations merged, and you shouldn't be surprised if they give up; anyway, your loss: I am not the one going to have an eternally half translated engine - in the best case.
Obviously you don't care that much, isn't it? So why should I? Enough said I guess.
 

User avatar
Wuzzy
Member
 
Posts: 3643
Joined: Mon Sep 24, 2012 15:01
GitHub: Wuzzy2
IRC: Wuzzy
In-game: Wuzzy

Re: Localization files not being used in Minetest releases

by Wuzzy » Sat Aug 10, 2019 05:48

FYI: Since 5.0.0, the translation chaos has settled. It seems that things got confused a lot when 0.4.17 came out, but that has been solved now. 0.4.17 was kind of a “special” release, but the way translations got handled for this release was a trainwreck. As far I understood, the devs just thought they had no new strings in 0.4.17, but they were wrong.

Anyway, this is now over. Since we have 5.0.0 now, Weblate is now used properly again. You can use Weblate safely again.

But what absolutely baffles me how the core devs were completely unable to admit they just completely screwed up translations in the 0.4.17 release. Instead, it's just laughable excuses all the way.

I mean, seriously, with this kind of attitude you don't need to be surprised when contributors jump ship.
My creations. I gladly accept bitcoins: 17fsUywHxeMHKG41UFfu34F1rAxZcrVoqH
 

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Re: Localization files not being used in Minetest releases

by Hamlet » Sat Aug 10, 2019 08:03

Wuzzy wrote:Anyway, this is now over. Since we have 5.0.0 now, Weblate is now used properly again. You can use Weblate safely again.


This is good news! :)
Thank you for telling me, I will start translating the localization file.
 

User avatar
Wuzzy
Member
 
Posts: 3643
Joined: Mon Sep 24, 2012 15:01
GitHub: Wuzzy2
IRC: Wuzzy
In-game: Wuzzy

Re: Localization files not being used in Minetest releases

by Wuzzy » Mon Aug 12, 2019 05:41

Note that when v5.1.0 is on the way, they will update Weblate again, so don't forget to come back. As far I know, they will only update Weblate very late, that is, shortly before release. There doesn't seem to be any Weblate updates mid-development.

But you can still start translating now, what you translate now will still carry over to the next version, it just probably won't be 100% then.
My creations. I gladly accept bitcoins: 17fsUywHxeMHKG41UFfu34F1rAxZcrVoqH
 

User avatar
Hamlet
Member
 
Posts: 661
Joined: Sat Jul 29, 2017 21:09
Location: Lombardy, Italy
GitHub: h4ml3t
IRC: H4mlet

Re: Localization files not being used in Minetest releases

by Hamlet » Thu Aug 15, 2019 16:39

Wuzzy wrote:As far I know, they will only update Weblate very late, that is, shortly before release. There doesn't seem to be any Weblate updates mid-development.


How much shortly? Are we talking about hours or days?
Is the Weblate update announced somewhere?
 

User avatar
Linuxdirk
Member
 
Posts: 2226
Joined: Wed Sep 17, 2014 11:21
Location: Germany
In-game: Linuxdirk

Re: Localization files not being used in Minetest releases

by Linuxdirk » Thu Aug 15, 2019 17:16

Hamlet wrote:Are we talking about hours or days?

When closely following Minetest’s development “shortly” is weeks or months.
 

User avatar
paramat
Developer
 
Posts: 3483
Joined: Sun Oct 28, 2012 00:05
Location: UK
GitHub: paramat
IRC: paramat

Re: Localization files not being used in Minetest releases

by paramat » Thu Oct 03, 2019 21:45

See Krock's post above and note:
Krock wrote:... 5.0.0 is developed on the master branch and 0.4.17(.1) is only a backport, which happened on a separate branch, which cannot be translated by Weblate (limited to master).

So translations unfortunately could not be used in 0.4.17. The accusations made here by Hamlet and Wuzzy about core developers are childish, pathetic and ridiculous.
 

User avatar
Wuzzy
Member
 
Posts: 3643
Joined: Mon Sep 24, 2012 15:01
GitHub: Wuzzy2
IRC: Wuzzy
In-game: Wuzzy

Post-mortem of 0.4.17

by Wuzzy » Fri Oct 04, 2019 02:35

I will use this post to write a post-mortem of 0.4.17's translations and recap about what happened, and what exactly went wrong, and what I think should have been done instead. It's time to have a more sober look on this.

I honestly think that the way translations were handled in 0.4.17 was awful, for several reasons. Mistakes were made, but I think a lot of useful lessons can be learned from this.

First, a recap of what has happened:

  1. 0.4.17 was released, one year after 0.4.16. 0.4.17 was generally considered a “bugfix release”
  2. Hamlet opened this thread, noting the untranslated strings in 0.4.17
  3. Krock said that Weblate cannot be used to translate for 0.4.17 because Weblate doesn't support it somehow
  4. In other words, 0.4.17 broke translations with no way for translators to update for that version
  5. According to the early posts in this thread, the drop in coverage was significant: From 100% to 77%. So the best a translator could do for 0.4.17 was 77% coverage.
  6. Ironically, this meant that the “bugfix release” had a major regression, namely, 23% of strings became untranslated, guaranteed.
  7. When core devs were asked for other ways to contribute translations for 0.4.17, the answer was that all alternatives (like PRs) were basically shut down. It was impossible to get 100% for any translation in 0.4.17
  8. Unsurprisingly, Hamlet was really upset when learning about this and threatened to jump ship. Several reasons: Hamlet completed a translation 100%. 1 year difference between 0.4.16 and 0.4.17, so significant waiting time. Major drop in translation coverage in 0.4.17. When the core devs were asked to explain, Hamlet learned that all work went to waste for 0.4.17, with no intention to fix. All this feeds into frustration.
  9. Core devs promised that all will be well in 5.0.0 again.
  10. The promise became true: In version 5.0.0, translations are handled properly again; the chaos has (finally) settled
  11. No core dev has admitted any mistake from their side yet, despite being requested twice

To summarize, translations for 0.4.17 were a complete trainwreck, but the train got repaired and got back on track in 5.0.0.

... 5.0.0 is developed on the master branch and 0.4.17(.1) is only a backport, which happened on a separate branch, which cannot be translated by Weblate (limited to master).

Well, of course I read that, and I understand that. So this will go into consideration.

And yet I stand by my claim that this was a screw-up. I don't mean this as an insult. We all screw up sometimes. But what's worse than screwing up is covering up your mistakes. This is the reason why I mentioned number 11 above. It's not about shaming people. It's about acknowleding when something got wrong, so we can move on and work on solutions.

It's also a screw-up in terms of management. There are several reasons why it's a screw-up:

  1. My guess is that core devs simply believed that 0.4.17 had no string changes, so they simply copied 0.4.16 strings over. But they were wrong (it was always called a “bugfix release”, after all). And when string changes were noticed, it was way too late. This would explain a lot.
  2. Weblate apparently does not support any branch except master. If that's true, then this is a screw-up of Weblate. I don't know if this is actually true. But it really sounds like an annoying restriction to me.
  3. A “bugfix release” broke 23% of translated string with no way to update, so it introduced a regression
  4. It was implied there were no alternatives, but I can think of one: Since there was (apparently) no simple solution for making Weblate behave, one possible solution to avoid the mess would have been to simply revert all string changes in 0.4.17 to avoid breaking existing translations
  5. None of this was communicated clearly in advance. It only was uncovered after Hamlet complained. (But it could be because of screw-up reason 1, however). This meant, since the discussion was missing, no possible solutions or alternatives were able to be discussed

Lessons learned:
  1. Never assume that zero strings have changed, even for so-called “bugfix releases”
  2. For Weblate, there's no easy lesson to learn, I'm afraid. :-( Maybe investigate whether this restriction really applies, or if there's a workaround somehow?
  3. It should be forbidden to break translations for a future release if its impossible to translate it 100%
  4. If you break translations like that (knowingly or not), you WILL piss off translators and risk of shying them away for good
  5. Communicate breaking changes before release, for example, in the News forum, to alleviate potential breakage or even find a real solution
  6. Backport releases are a mess. Maybe don't do backport releases? :D But if you must do them, be extra careful

So, while now, with MT 5.0.0, everything is fine and dandy again, I urge to not forget the chaos of 0.4.17. If the next time such a “backport” release comes along (e.g. for Minetest 6.0.0), I hope the same mistakes will not be repeated, and the lessons will be applied.

Thank you for reading this overlong post.

End of post-mortem.
My creations. I gladly accept bitcoins: 17fsUywHxeMHKG41UFfu34F1rAxZcrVoqH
 


Return to Problems



Who is online

Users browsing this forum: No registered users and 1 guest