Le ven. 27 nov. 2020 à 13:44, andy pugh <bodge...@gmail.com> a écrit :

>
> That sounds promising. What happens next?
>

If there's no problem for everyone at this stage, I'd like to push the
Crowdin test, together with POedit, as well as OmegaT, up to what could be
considered an alpha setup.
All documentation files should be imported at this stage. Translation files
are still to be uploaded and a proper way to manage PO files remains to be
found without a reference english POT file...
Then, have to get a clearer understanding of branches management and usage.
Finally bring others to test that alpha stage and support them doing so.

Of course, anyone interested is more than welcome to jump in at anytime.
Just join me at https://crowdin.com/project/linuxcnc !


I am a little nervous about committing to an external tool, LinuxCNC has
> always been rather self-contained. (We used to host our own git repo too,
> Github is fairly recent)
>

Believe me, I'm too! But, considering the statements of the company, the
number of Open Source projects managing their translations on Crowdin
(among them FreeCAD), and all the more the fact that all datas are fully
downloadable in standard formats supported by other self hosted
platforms... I'm keen to consider it as an acceptable option.

Like I said, I will anyway have a look at Weblate and Pootle. But if they
appear to be as promising, then we'll have to deal with hosting...


Still have to test it further, but we may be able to translate man pages
> > directly using the tools already mentionned.
>
> That would potentially be useful for the hand-written manpages.
> The ones generated by halcompile  are probably best translated in the
> source as their manpage entries are generated fresh each time.
> (Though that would all depend on how the translation service detects
> updates)
>

Well, as much as I value documentation as code, and from my current
understanding, I tend to think that code should only embed english
language. English translation (pot), documentation and man pages should be
(automatically via GH hooks?) generated from these strings. Translation
toolchain, integrated with repository, should be built upon those reference
language files. Translation platform tracs reference language updates as
well as translations ones with regards to english.

Translation can take place on web platform using all the really fancy
features proposed (glossary, machine translation, translation memory,
duplicate strings management, context and screenshots...), on desktop with
graphical tools such as POedit or OmegaT, with text editors using their
respective PO plugins/modes/extensions.
Aside from POedit which connects directly to Crowndin (still have to test
this), translations made "offline" can be uploaded via the web. Also have
to see if repository sync allows to push translations directly and how
conflicts are handled in such case. I suspect that this is left at the
git/repository level with crowdin only pushing translation updates.

Let me know what yall think. This is very knew to me so any constructive
opinion, comments and furthermore testing is of course more than welcome.

Thanks in advance 🙏

_______________________________________________
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to