Page 1 of 1

Replacing Transifex

PostPosted: 21 Aug 2018, 13:37
by Wuzzy
I noticed that lots of games are putting the locale files on Transifex for translations. Even projects on FreeGameDev are on Transifex, like SuperTuxKart and recenly Me & My Shadow.

I am worried we are becoming increasingly dependent on a single company which might screw us all over at any time. But that's not the biggest issue, since translation files are usually frequently backed up.

The biggest problem with Transifex is the JavaScript. We have all fallen into the JavaScript trap. The fact is, anyone who is using Transifex is using and encouraging the use of proprietary software, namely, the JavaScript. If you look into the code, you will see that it's actually obfuscated and impossible to decypher. And without JavaScript the website is just broken.

I am also part of it, sadly, I took the bait, hook, line and sinker. You might call me a hypocrite now, but that doesn't change that the problem is real or that I or you cannot change that guilty habit.

I think this is (sadly) another classic case of “convenience trumps freedom”. What can we do (in general) to move away from this dependency?

Some ideas:
- Convince Atlassian to release their JavaScripts as free software (hahahahahaha!)
- Move to
- Develop a free standalone client for Transifex, using their API. At least that would fix the JavaScript trap but not the dependency on a proprietary software vendor. Could be frustrating when the company either changes the API extremely or shuts it down
- Launch and host a competing community-driven platform based on Pootle, aimed specifically for free software games. Which has similar features and appears to be free and concinve projects to move. Similar to

I believe using would work for most projects. Weblate supports Gettext (PO), and Qt .ts files and a couple more “esoteric” file formats ( But it doesn't support any weird hand-hacked format, of course. So, simply moving to seems to be sensible option, too. I used it before to translate Minetest and it works just fine, just without the blobs.

I believe the 3rd solution would be best for projects, if Weblate supports your format. Your project should probably be using Gettext anyway. :)

I think the reason why Transifex is so popular because everyone else is using it, and for translators this means the find many projects on one place. But this comes at a price …
In general, the idea here is that having a competing platform specifically for free software projects and also being free software itself might also bind the community a little bit closer together which is of course good in itself. :)

There's also Launchpad in theory but I hate Launchpad (at least the instance hosted by Canonical) because it's extremely slow and takes ages to load. I wouldn't recommend it, esp. we have better free alternatives. Cataclysm: Dark Days Ahead once was on Launchpad and actually moved away because of this.

List of free game projects currently on Transifex (excerpt):
- SuperTuxKart
- Me and my Shadow (currently experimental, it seems)
- SuperTux
- Flare
- The Butterfly Effect
- Cataclysm: Dark Days Ahead
- Teeworlds
- Stunt Rally
- 0 A.D.
- Voxelands
- Widelands
- Neverball
- Pioneer
- Xonotic

(Possibly incomplete) List of free game projects on (is itself free software):
- Minetest and related projects
- Summoning Wars
- Godot Engine
- Andor's Trail

This is just a rough overview, but I think there's currently a strong preference on Transifex and an obvious discrepancy here. A lot of the projects listed above are Gettext-based and thus could already be moved to

The irony that apparently a majority of free software games have made themselves dependent on a proprietary vendor is not lost on me.

Related topic: (Online translation platforms)

EDIT: It turns out Weblate supports more formats than just PO, sorry for the mistake! See here for the full list:

Re: Replacing Transifex

PostPosted: 21 Aug 2018, 16:11
by acme_pjz
The requires you to send an application letter for translation hosting, which is inconvenient. I think it's a reason why few people are using

In fact, there are another weird point on Transifex: it requires your FOSS project are non-commercial, i.e. not funded, etc. This makes it infeasible for projects like Godot Engine.

My suggestion is hosting a pootle or weblate on freegamedev, since there are discussions on hosting a gitea on freegamedev --- why don't we have both of them?

Re: Replacing Transifex

PostPosted: 21 Aug 2018, 18:08
by Julius
acme_pjz {l Wrote}:My suggestion is hosting a pootle or weblate on freegamedev, since there are discussions on hosting a gitea on freegamedev --- why don't we have both of them?

True :)

I will look into that soonish.

Edit: Hmm, both are Python based. Does anyone have some experience with the server requirements? Also, is there a preference for weblate or pootle?
Edit2: Weblate has a convenient official Docker image up on Dockerhub, so I am probably going to test that first, unless there are strong objections. But probably tomorrow or so.

Re: Replacing Transifex

PostPosted: 22 Aug 2018, 12:10
by GunChleoc
I'm a Pootle fan, so i'd love to see one hosted. I have been meaning to create an instance for Battle for Wesnoth for over a year now because they still have submissions per e-mail, but I have been unable to find the time to learn. AFAIK it is based on Django.

The reason everybody uses Transifex is because it works and they are always open to feedback.

Re: Replacing Transifex

PostPosted: 22 Aug 2018, 19:17
by Wuzzy
I will look into that soonish.

That would be awesome!

I have no strong preference to either Weblate or Pootle. I only had experience with Weblate so far, I can't say whether Pootle is better. Just from the feature set, Pootle seems to be more general-purpose.

why don't we have both of them?

I am also in favor of a Gitea instance. :)

Re: Replacing Transifex

PostPosted: 24 Aug 2018, 19:15
by Julius
Sorry for the delay, but I realized that I first need to (finally) configure an email system that allows sending mail through SMTP. This is a bit complicated due to our hosting setup. But without it quite a few web-apps can't be installed (like weblate) and I don't like using a gmail account or similar for that.
I have it half working, so it is just a matter of fixing it a bit more I hope ;)

About Gitea: not before they implement a per user quota on the repository size... and even then I probably need to increase my hosting's SSD drive space.

Re: Replacing Transifex

PostPosted: 25 Aug 2018, 13:40
by GunChleoc
Good luck and thanks for all your work on the site!

Re: Replacing Transifex

PostPosted: 11 Sep 2018, 17:53
by Julius
Sorry, I put this on the back-burner for now. Too busy with other stuff. I hope no-one was actually waiting for this. Maybe towards the end of the year I will revisit this.

Re: Replacing Transifex

PostPosted: 10 Nov 2018, 16:47
by shirish
just putting it so it's on my radar as well, if and when this is implemented.

Weblate supports other formats, too

PostPosted: 25 Jan 2019, 01:48
by Wuzzy
I like to correct an error of mine.

I have claimed that Weblate only supports the Gettext PO format. But this is not true. I have looked in the docs again and noticed Weblate supports a variety of formats:

The most striking format in the list is the support of the “.ts” file format used by Qt, which is quite important (besides PO). That's quite a relief! I can't tell much about the other formats, I never used or needed them.

Weblate doesn't seem to support completely arbitrary file formats that you made up on the spot, but then your project might have other problems, too. ;)

I don't know where I got the idea that Weblate is Gettext-only. Silly me.


Setting freedom aside, what do you think about Weblate in general?

I think Weblate has also a nice interface. It's nice and simple and gets the job done. No distractions, no bullshit. You can download the files whenever you want. Glossary support … It doesn't get in the way of your work. Basically all the basic features you'd expect such a tool to have.
So, I think, Weblate is a complete legit tool, I really don't understand the current skepticism towards it. But your milage may vary.

Re: Replacing Transifex

PostPosted: 07 Feb 2019, 01:27
by mdtrooper
The people from Framasoft (it is a famous french colective of free software) uses Zanata. And this is their online instance.

Re: Replacing Transifex

PostPosted: 07 Feb 2019, 12:04
by NaN
I am not sure I understand the issue here.

Transifex is just a tool. Is it convenient? Yes. Is it necessary? Absolutely not.

You/translators own the translations. Just like with git you have a local clone. You can work on the translation online. But I've always preferred to do it locally. So my interactions with Transifex have been limited to pushing/pulling translations, managing translators. This could be done in a forum as well, it is just a bit less convenient.

Re: Replacing Transifex

PostPosted: 08 Feb 2019, 23:30
by Lyberta
NaN {l Wrote}:I am not sure I understand the issue here.

Transifex is just a tool. Is it convenient? Yes. Is it necessary? Absolutely not.

Then why do people screw other people over with unnecessary proprietary tools? That's the issue.

Re: Replacing Transifex

PostPosted: 09 Feb 2019, 19:25
by GunChleoc
Well, what the convenience of the "unnecessary" tool brings to me as a translator is that I can spend my valuable time translating rather than having additional management overhead by manually checking everything. Also, automatic e-mail notifications are an absolute must when you're maintaining 20+ projects. Watching the repos themselves doesn't cut it, because it would totally spam my inbox with code changes that are completely irrelevant to translators.

So: Is it necessary? Absolutely yes. It doesn't have to be Transifex, but it has to be something that will give me an overview of the stats, easy downloading/uploading, and e-mail notifications. If any of these features is missing, our time will be significantly wasted. Until recently, there were only 2 localizers doing volunteer work for my locale, and we have been recently joined by a 3rd. So, our time is a very scarce commodity and we have to pick our battles ;)

I think using FLOSS as much as possible is important, but I can't afford to take an absolute stance on this.

Re: Replacing Transifex

PostPosted: 21 May 2019, 12:22
by Julius
Revisiting this now. Weblate seems to be the best solution right now as it can be linked up to a git repository for automation.

What concerns me a bit is that the website claims it needs 2GB of RAM on the server. Might be a case of making sure no-one tries to run this on a RPI1, but if true this would be by far the most resource intensive new service. Its a Python project, so that is possible (I try to stay away from Python and Ruby stuff normally due to high RAM demand).

Sadly the Golang based translation project recently ceased development and never got very far comparatively.

Re: Replacing Transifex

PostPosted: 22 May 2019, 22:51
by Wuzzy
I'm excited to read about future events!

Re: Replacing Transifex

PostPosted: 24 May 2019, 23:39
by Julius
Hmm... spend some time trying to get Weblate to work. The basic setup is easy, but more advanced stuff and specifically one that fits to my specific server setup sadly not. So for now (unless I come up with a better idea) this is not realistic :(

The alternative seems to be Pootle with Serge:

There is a promising Docker recipe that on first glance would fit well to my setup, but it's late now and I will try it another day:

Anyone has any experience with this kind of setup?

Re: Replacing Transifex

PostPosted: 03 Dec 2019, 10:07
by GunChleoc
Framasoft now have a weblate instance:

Peertube has moved there from Zanata and I'm glad they did; I didn't really get on with the Zanata interface.