Re: translation infrastructure in need

This is an OpenPGP/MIME signed message (RFC 4880 and 3156)
--===============4879980765134213834==
Content-Type: multipart/signed; protocol="application/pgp-signature";
 micalg=pgp-sha512;
 boundary="------c169939cccba5fb9955863314da9a623b045f49da7cad50bd9d35a0819e85cf7";
 charset=utf-8

This is an OpenPGP/MIME signed message (RFC 4880 and 3156)
--------c169939cccba5fb9955863314da9a623b045f49da7cad50bd9d35a0819e85cf7
Content-Type: multipart/mixed;
	boundary="18b1894b3dc582fd_ee4e3dd7855b7e03_26ed2bddbe0d5fb7"


--18b1894b3dc582fd_ee4e3dd7855b7e03_26ed2bddbe0d5fb7
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable

Hey all,

I have been working in this for the last little while, although I confess I =
have not been so good at keeping those efforts public. My efforts are too un=
stable to review meaningfully.

Thank you for the suggestions and help Marco.

I have here a quick update on progress. Its my first time working with po4a =
(or po files) so it took me some time to get an understanding of the situati=
on. That understanding is still incomplete.

Just keeping you posted on progress
- Our initial idea of one po file per priority does not really work. It trie=
s to deduplicate text and translations across files, so each paragraph has t=
o be translated identically across all files in that priority. That seemed n=
ot ideal, so now I think one po per file, as seems to be more standard.  Web=
late has a priority mechanism, so we can try use that, however it works.
- Any difference in number / order of XML elements whatsoever causes po4a to=
 fail to salvage text from translations. This is the case for a huge number =
of our translated files. For an overview, that is at present _less_ strict t=
han po4a (I will patch it to match soon) see https://status.fsfe.org/xml_str=
ucture/ . We introduced ci checking for such structural differences some tim=
e ago to work towards resolving this issue.
- po4a is designed for an initial scrape of all text, and then keeping the p=
o files as the source of truth. We have many files that it cannot scrape, as=
 explained above. This presents an issue. Let us say we have an English mast=
er, and Polish translation with an XML difference. Po4a cannot salvage the p=
olish translation, and hence considers it does not exist. And so it seems to=
 have a tendency to delete it when updating the source files from po, as it =
considers it untranslated. This seems annoying to avoid, but I have not inve=
stigated it so much with per file po files. Perhaps that will help.

Hope this clarifies the situation a little.

I am now entering my university exams, so I will be busy for some time, and =
unable to work on this.

Yours,
Darragh=20
-------- Original Message --------
On Thursday, 05/21/26 at 10:45 Nico Rikken <[email protected]> wrote:
Hi Marco,

Thanks for the details. It seems you already tested to verify it works. I pl=
an to set aside some time in the coming weeks to test it for myself.

I expect we can avoid back and forth by involving more people including Darr=
agh and system-hackers early on and to take it step by step. As said there h=
ave been earlier attempts, so I don't expect unwillingness. It of course wil=
l have to fit into the larger system for translating and building the websit=
e.

I hope this will get us past the hurdle.

Best,
Nico
_______________________________________________
Translators mailing list -- [email protected]
To unsubscribe send an email to [email protected]

This mailing list is covered by the FSFE's Code of Conduct. All
participants are kindly asked to be excellent to each other:
https://fsfe.org/about/codeofconduct
--18b1894b3dc582fd_ee4e3dd7855b7e03_26ed2bddbe0d5fb7
Content-Type: application/pgp-keys;
	filename="publickey - [email protected] - 0x2F56B4A2.asc";
	name="publickey - [email protected] - 0x2F56B4A2.asc"
Content-Disposition: attachment;
	filename="publickey - [email protected] - 0x2F56B4A2.asc";
	name="publickey - [email protected] - 0x2F56B4A2.asc"
Content-Transfer-Encoding: base64
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--18b1894b3dc582fd_ee4e3dd7855b7e03_26ed2bddbe0d5fb7--

--------c169939cccba5fb9955863314da9a623b045f49da7cad50bd9d35a0819e85cf7
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: ProtonMail

wrsEARYKAG0FgmoOyqoJEFRZ86rSLtKARRQAAAAAABwAIHNhbHRAbm90YXRp
b25zLm9wZW5wZ3Bqcy5vcmcMB9X3Hd7rvcvHzl1dVHmXJO3mNMO1aGrG0yAt
j6n6/BYhBC9WtKJ2/VTp9/QoTlRZ86rSLtKAAABnOwEAqVW9huDwtGa/P9Wv
WWGWlpRUlgga7E21r+8hQf8ZCdwBAPniFRa+PaLMCbBC2SoilFttSf3M1QVP
cVv4gz1BcxYH
=zMkH
-----END PGP SIGNATURE-----


--------c169939cccba5fb9955863314da9a623b045f49da7cad50bd9d35a0819e85cf7--


--===============4879980765134213834==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Translators mailing list -- [email protected]
To unsubscribe send an email to [email protected]

This mailing list is covered by the FSFE's Code of Conduct. All
participants are kindly asked to be excellent to each other:
https://fsfe.org/about/codeofconduct
--===============4879980765134213834==--