#1131786 hunspell-sv: please update to 2.42+ with 269,221 words

#1131786#5
Date:
2026-03-24 14:23:02 UTC
From:
To:
The hunspell-sv package in Debian ships version 1.51 of the Swedish
dictionary, while upstream (SFOL / Den Stora Fria Ordlistan) has been
at version 2.42 for some time.

Additionally, we have significantly expanded the dictionary with words
from two academic sources:

- SALDO (Språkbanken, University of Gothenburg): +17,000 lemmas (CC-BY 3.0)
- Folkets Lexikon (KTH): +900 curated words (CC-BY-SA 2.5)

The new dictionary contains 269,221 base words (up from ~154,000 in 1.51
and ~251,000 in 2.42), making it comparable to commercial spell checkers.

A Gerrit patch for the LibreOffice dictionaries repository (which Debian
uses as upstream) is available:
https://gerrit.libreoffice.org/c/dictionaries/+/202383

The updated dictionary is also available on GitHub:
https://github.com/yeager/hunspell-sv

Would it be possible to update the package to the latest upstream?

Thank you,
Daniel Nylander

#1131786#10
Date:
2026-03-24 17:32:57 UTC
From:
To:
notfound 1131786 1.51-3

tag 1131786 + moreinfo

retitle 1131786 hunspell-sv: please update to 2.42+ with 269,221 words

thanks


Hi,

Am 24.03.26 um 15:23 schrieb bosse@danielnylander.se:

I don't see how you got this version at all. Besides the fact that this is built out of libreoffice-dictionaries and doesn't get special handling and thus is

$ grep version description.xml
<?xml version="1.0" encoding="UTF-8"?>
     <version value="2.40" />
         <OpenOffice.org-minimal-version value="3.0" d:name="OpenOffice.org 3.0" />

this is even in stable:

rene@frodo:~/libreoffice-dictionaries-25.2.3/dictionaries/sv_SE$ grep version description.xml
<?xml version="1.0" encoding="UTF-8"?>
     <version value="2.40" />
         <OpenOffice.org-minimal-version value="3.0" d:name="OpenOffice.org 3.0" />

lbreoffice-dictionaries 7.5.0 from oldstable doesn't even contain a version.

Wrong.

$ grep version description.xml
<?xml version="1.0" encoding="UTF-8"?>
     <version value="2.40" />
         <OpenOffice.org-minimal-version value="3.0" d:name="OpenOffice.org 3.0" />
Which is not merged. So your subject a) is wrong from the version you come from and b) from the version you are wanting me to go to.

GitHub is not  the source of this package. :)

As shown, the package is on par with what is current in libreoffice-dictionaries' repo.


Ideally someone would package it up from GitHub and then it can be removed here.


Regards,


Rene

#1131786#21
Date:
2026-03-24 17:50:11 UTC
From:
To:
Hi Rene,

Thanks for the correction — you're right, I apologize for the wrong
version number. The 1.51 was from an old Ubuntu installation that
apparently still had a very old package.

You're also correct that the Gerrit patch needs to be merged first.
We'll follow up on that with the LO maintainers.

Your suggestion to package it separately from GitHub is a good one —
we already have a DEB package at https://github.com/yeager/debian-repo
(hunspell-sv 2.42.3 with 279,121 words). Would it make sense to
submit this as a separate Debian package (e.g., hunspell-sv-extended)
that doesn't conflict with the libreoffice-dictionaries version?

Regards,
Daniel Nylander

#1131786#26
Date:
2026-03-24 18:01:07 UTC
From:
To:
Hi,

Am 24.03.26 um 18:50 schrieb bosse@danielnylander.se:

No. You (or one) could replace hunspell-sv without any problem (version as 2:2.42-1 or so) and be done and I can drop it.


Regards,


Rene

#1131786#31
Date:
2026-03-25 05:38:31 UTC
From:
To:
Hi Rene,

Thanks for the quick response and the corrections.

You're right — the version in description.xml is 2.40, not 1.51. My
apologies for the incorrect version reference. I retested and the 1.51
came from a stale dpkg query on a bookworm system that hadn't been
updated.

The core request remains: the current dictionary in
libreoffice-dictionaries has ~154,000 words (SFOL 2.42 base), and
we've expanded it to 269,221 words by integrating additional
freely-licensed sources (SALDO from Språkbanken/University of
Gothenburg, CC-BY 3.0, and Folkets Lexikon from KTH, CC-BY-SA 2.5).

The Gerrit patch
(https://gerrit.libreoffice.org/c/dictionaries/+/202383) is pending
review upstream. Once merged, it would flow into Debian naturally.

Your suggestion to package hunspell-sv separately from GitHub is also
interesting — it would allow more frequent updates independent of the
libreoffice-dictionaries release cycle. We'd be happy to help with
packaging if someone picks that up.

Best regards,
Bosse Nylander