Translation project maintainers,
We, Fedora Weblate administrators, are considering enabling the Weblate's "Contributor in comment"[1] add-on to the whole Fedora Weblate instance. It updates the comment part of the PO file header to include contributor names and years of contributions.
This is planned for March 6, 2026 if nothing comes up. See the tracker issue at localization/tickets #51 [2].
Please reach us if you have any questions or objections.
Best regards, Rafael
[1] https://docs.weblate.org/en/weblate-5.16.1/admin/addons.html#contributors-in... [2] https://forge.fedoraproject.org/localization/tickets/issues/51
On Fri, Feb 27, 2026 at 09:46:28PM -0300, Rafael Fontenelle via trans wrote:
Translation project maintainers,
We, Fedora Weblate administrators, are considering enabling the Weblate's "Contributor in comment"[1] add-on to the whole Fedora Weblate instance. It updates the comment part of the PO file header to include contributor names and years of contributions.
This is planned for March 6, 2026 if nothing comes up. See the tracker issue at localization/tickets #51 [2].
Please reach us if you have any questions or objections.
That ticket is missing a compelling rationale for this change, as IMHO listing author names in the .po files (or any source file) is an obsolete hangover from the days before people used reliable version control for change tracking.
In upstream projects I'm maintaining that use Fedora weblate, we long ago stopped recording any "Author" lines in most source code files, with the intent that every contributor is accurately identified by the git commit author records. We have weblate configured to merge commits "by author", so we can retain a record of what work each translator did.
IMHO "Contributor in commit" plugin should remain optional for any projects which are preserving translator authorship via git history, rather than force this obsolete history tracking concept on all projects.
With regards, Daniel
Daniel P. Berrangé via trans píše v po 02. 03. 2026 v 09:14 +0000:
On Fri, Feb 27, 2026 at 09:46:28PM -0300, Rafael Fontenelle via trans wrote:
Translation project maintainers,
We, Fedora Weblate administrators, are considering enabling the Weblate's "Contributor in comment"[1] add-on to the whole Fedora Weblate instance. It updates the comment part of the PO file header to include contributor names and years of contributions.
This is planned for March 6, 2026 if nothing comes up. See the tracker issue at localization/tickets #51 [2].
Please reach us if you have any questions or objections.
That ticket is missing a compelling rationale for this change, as IMHO listing author names in the .po files (or any source file) is an obsolete hangover from the days before people used reliable version control for change tracking.
In upstream projects I'm maintaining that use Fedora weblate, we long ago stopped recording any "Author" lines in most source code files, with the intent that every contributor is accurately identified by the git commit author records. We have weblate configured to merge commits "by author", so we can retain a record of what work each translator did.
IMHO "Contributor in commit" plugin should remain optional for any projects which are preserving translator authorship via git history, rather than force this obsolete history tracking concept on all projects.
With regards, Daniel --
Our motivation is encouragement of translators credits. As a translation community we have perhaps a bit closer to .po files than git(s). Though listing translators in .po files on its own is not guarantee of the credit, neither is it the commit authorship.
My view is your concern is reasonably understandable and we have already touched that in our internal discussion. Nevertheless, I still tend to think that the projects sensitive to the space/size requirements are smaller in quantity.
And yes, the 'Contributors in comment' add-on stays optional in an opt-out way. It deserves inclusion either in our documentation and 'request template'.
With regards,
Josef
Libvirt is a good example of what upstream project should do: properly identify contributors in git log: https://github.com/libvirt/libvirt/commits/master/po
I wonder which part of the 594 contributors listed in github are translators.
I assume if there is a credits somewhere in the software, it also lists contributors.
To my knowledge, libvirt is an exception, and translator credits is often missing, which may lead translators to feel like second-zone contributors.
I do prefer to have a simple header to read a contributors, because: * translators copy/share po files to for translation memories purposes (which removes the git history) * git knowledge should not be required to get it touch with other translators
But this is a personal belief, and we should not impose our beliefs.
Having the "contributor in comment" plugin enabled by default on the Fedora translation platform is fine to me, as long as we inform project maintainers, and allow them to disable it if they have another clean way to properly identify contributors.
The challenging question I see is: what do we do if there is no translator credits either in git log or file header?
On Sat, Mar 07, 2026 at 08:38:23AM +0100, Jean-Baptiste via trans wrote:
Libvirt is a good example of what upstream project should do: properly identify contributors in git log: https://github.com/libvirt/libvirt/commits/master/po
I wonder which part of the 594 contributors listed in github are translators.
I count approx 90 contributors to *.po files in libvirt.git.
I assume if there is a credits somewhere in the software, it also lists contributors.
The source tarballs also contain an AUTHORS.rst file whose content is generated from git histort at time of release.
To my knowledge, libvirt is an exception, and translator credits is often missing, which may lead translators to feel like second-zone contributors.
Absolutely, that is not acceptable. Translators must be attributed to the same extent that any other contributors to the project are. This is one of aspects I really like about the combination of Weblate with git, and the git forges. It integrates translators work into the same mechanism that all other contributions arrive, so everyone is treating equally.
I do prefer to have a simple header to read a contributors, because:
- translators copy/share po files to for translation memories purposes
(which removes the git history)
- git knowledge should not be required to get it touch with other
translators
One of the reasons we use Weblate for upstream libvirt is that Weblate provides the platform for community interaction and sharing of translated strings in a way that copying around .po files can never match.
Occassionally we get people coming directly to upstream wnating to translate stuff, but we tend to ask them to join to the Fedora weblate community. Only very rarely would we accept fixes directly to the .po files and submit them back to weblate ourselves.
But this is a personal belief, and we should not impose our beliefs.
Having the "contributor in comment" plugin enabled by default on the Fedora translation platform is fine to me, as long as we inform project maintainers, and allow them to disable it if they have another clean way to properly identify contributors.
Having the plugin enabled by default is OK with me from a libvirt POV, provided we retain the ability to opt-out of this change per project.
The challenging question I see is: what do we do if there is no translator credits either in git log or file header?
I would file an issue against the project requesting that they fix it to credit translators, in the same manner that they are creditting any other translators. Consistency is the key IMHO. There are a variety of approaches for creditting contributors, and whichever approach a project decides upon, should be applied equally to contributors or any type of content.
With regards, Daniel
-
Daniel P. Berrangé -
Jean-Baptiste -
Josef Hruska -
Rafael Fontenelle