apertium-stuff — Questions related to apertium project

Re: [Apertium-stuff] HFST and lttoolbox languages in translation pairs

[Daniel S. <awe...@gm...>]

Hello and welcome back!

When you set up your new pair using aperium-init, you'll need to specify
which analyzers the two monolingual modules use and it will take care of
the details, which will only matter later if you need to do something
complicated in the makefile.

As it happens, I also expect to be dealing with grc-eng in the near future,
so we should discuss further.

Daniel

On Sun, May 25, 2025 at 1:20 PM Samuel Sloniker <sa...@kj...> wrote:

> Hello! My name is Samuel Sloniker. I previously participated in the
> Apertium project during and for a while after Google Code-in 2019 under the
> name "scoopgracie." (I apologize for any inconvenience or disruption I
> caused during that time.) I have not been involved for a few years, but I
> am interested in creating a translation pair from Ancient Greek (primarily
> focusing on New Testament Koine) to English. I have noticed that
> `apertium-grc` uses HFST, whereas `apertium-eng` uses lttoolbox. Will that
> pose a problem in creating the pair, or can I just follow
> https://wiki.apertium.org/wiki/Apertium_New_Language_Pair_HOWTO, with the
> exception of not needing to create the monodixes?
>
> Samuel Sloniker
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>

[Apertium-stuff] HFST and lttoolbox languages in translation pairs

[Samuel S. <sa...@kj...>]

Hello! My name is Samuel Sloniker. I previously participated in the Apertium project during and for a while after Google Code-in 2019 under the name "scoopgracie." (I apologize for any inconvenience or disruption I caused during that time.) I have not been involved for a few years, but I am interested in creating a translation pair from Ancient Greek (primarily focusing on New Testament Koine) to English. I have noticed that `apertium-grc` uses HFST, whereas `apertium-eng` uses lttoolbox. Will that pose a problem in creating the pair, or can I just follow https://wiki.apertium.org/wiki/Apertium_New_Language_Pair_HOWTO, with the exception of not needing to create the monodixes?

Samuel Sloniker

Re: [Apertium-stuff] DOCX format handling on apertium.org and beta.apertium.org

[Marc R. I. <mar...@gm...>]

It's Transfuse indeed, unfortunately in the wrong way (the results are 
good when it's not used). I'll open an issue on GitHub.

Thanks!

Marc Riera


El 20/4/25 a les 15:56, Tino Didriksen ha escrit:
> It should only depend on whether Transfuse is installed and enabled. 
> Bad results are hopefully when Transfuse is not used - and if not, 
> then that's a bug in Transfuse.
>
> -- Tino Didriksen
>
>
> On Sun, 20 Apr 2025 at 15:46, Marc Riera Irigoyen 
> <mar...@gm...> wrote:
>
>     Thank you, Tino, I didn't know about that repository.
>
>     I really need to investigate this, because I just found out the
>     following:
>
>       * Windows Subsystem for Linux: with either stable or nightly
>         Apertium-core and Apertium-Apy, I get the same results as the
>         sites (good, consistent, even if the translation is worse).
>       * Arch Linux (original PC where I tested this): with
>         Apertium-core and Apertium-Apy compiled from source, for some
>         reason I get different results (bad, inconsistent). I would
>         expect to get the same as nightly.
>
>     I'll keep looking into this and hopefully report soon where the
>     issue is.
>
>     Regards,
>
>     Marc Riera
>
>
>     El 19/4/25 a les 16:46, Tino Didriksen ha escrit:
>>     The sites are built using
>>     https://github.com/apertium/apertium-website - beta is updated
>>     nightly, release is updated when there are new releases (which
>>     there were last week, but I forgot to run it until just now).
>>
>>     -- Tino Didriksen
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Re: [Apertium-stuff] DOCX format handling on apertium.org and beta.apertium.org

[Tino D. <ma...@ti...>]

It should only depend on whether Transfuse is installed and enabled. Bad
results are hopefully when Transfuse is not used - and if not, then that's
a bug in Transfuse.

-- Tino Didriksen


On Sun, 20 Apr 2025 at 15:46, Marc Riera Irigoyen <
mar...@gm...> wrote:

> Thank you, Tino, I didn't know about that repository.
>
> I really need to investigate this, because I just found out the following:
>
>    - Windows Subsystem for Linux: with either stable or nightly
>    Apertium-core and Apertium-Apy, I get the same results as the sites (good,
>    consistent, even if the translation is worse).
>    - Arch Linux (original PC where I tested this): with Apertium-core and
>    Apertium-Apy compiled from source, for some reason I get different results
>    (bad, inconsistent). I would expect to get the same as nightly.
>
> I'll keep looking into this and hopefully report soon where the issue is.
>
> Regards,
>
> Marc Riera
>
>
> El 19/4/25 a les 16:46, Tino Didriksen ha escrit:
>
> The sites are built using https://github.com/apertium/apertium-website -
> beta is updated nightly, release is updated when there are new releases
> (which there were last week, but I forgot to run it until just now).
>
> -- Tino Didriksen
>
>

Re: [Apertium-stuff] DOCX format handling on apertium.org and beta.apertium.org

[Marc R. I. <mar...@gm...>]

Thank you, Tino, I didn't know about that repository.

I really need to investigate this, because I just found out the following:

  * Windows Subsystem for Linux: with either stable or nightly
    Apertium-core and Apertium-Apy, I get the same results as the sites
    (good, consistent, even if the translation is worse).
  * Arch Linux (original PC where I tested this): with Apertium-core and
    Apertium-Apy compiled from source, for some reason I get different
    results (bad, inconsistent). I would expect to get the same as nightly.

I'll keep looking into this and hopefully report soon where the issue is.

Regards,

Marc Riera


El 19/4/25 a les 16:46, Tino Didriksen ha escrit:
> The sites are built using https://github.com/apertium/apertium-website 
> - beta is updated nightly, release is updated when there are new 
> releases (which there were last week, but I forgot to run it until 
> just now).
>
> -- Tino Didriksen
>
> On Sat, 19 Apr 2025 at 13:53, Marc Riera Irigoyen 
> <mar...@gm...> wrote:
>
>     Hello,
>
>     After being asked about DOCX documents with strange translations,
>     I ran a couple of tests and found out diverging results regarding
>     format handling. I have attached a minimal test document to
>     demonstrate the issue.
>
>     When translating from Spanish to Catalan, both apertium.org
>     <http://apertium.org> and beta.apertium.org
>     <http://beta.apertium.org> return broken translations (words are
>     split due to formatting and part of the sentence is missing).
>     However, running the latest versions of Apertium-apy and
>     Apertium-core locally with Spanish-Catalan pair compiled from
>     source gives better results: formatting is moved to word
>     boundaries and the whole sentence is translated.
>
>     Which versions of Apertium-core and Apertium-apy are being used on
>     the website at the moment?
>
>     Regards,
>
>     Marc Riera
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Re: [Apertium-stuff] DOCX format handling on apertium.org and beta.apertium.org

[Tino D. <ma...@ti...>]

The sites are built using https://github.com/apertium/apertium-website -
beta is updated nightly, release is updated when there are new releases
(which there were last week, but I forgot to run it until just now).

-- Tino Didriksen

On Sat, 19 Apr 2025 at 13:53, Marc Riera Irigoyen <
mar...@gm...> wrote:

> Hello,
>
> After being asked about DOCX documents with strange translations, I ran a
> couple of tests and found out diverging results regarding format handling.
> I have attached a minimal test document to demonstrate the issue.
>
> When translating from Spanish to Catalan, both apertium.org and
> beta.apertium.org return broken translations (words are split due to
> formatting and part of the sentence is missing). However, running the
> latest versions of Apertium-apy and Apertium-core locally with
> Spanish-Catalan pair compiled from source gives better results: formatting
> is moved to word boundaries and the whole sentence is translated.
>
> Which versions of Apertium-core and Apertium-apy are being used on the
> website at the moment?
>
> Regards,
>
> Marc Riera
>

[Apertium-stuff] DOCX format handling on apertium.org and beta.apertium.org

[Marc R. I. <mar...@gm...>]

Hello,

After being asked about DOCX documents with strange translations, I ran 
a couple of tests and found out diverging results regarding format 
handling. I have attached a minimal test document to demonstrate the issue.

When translating from Spanish to Catalan, both apertium.org and 
beta.apertium.org return broken translations (words are split due to 
formatting and part of the sentence is missing). However, running the 
latest versions of Apertium-apy and Apertium-core locally with 
Spanish-Catalan pair compiled from source gives better results: 
formatting is moved to word boundaries and the whole sentence is translated.

Which versions of Apertium-core and Apertium-apy are being used on the 
website at the moment?

Regards,

Marc Riera

Re: [Apertium-stuff] Permissions on repositories and mailing list

[Aure S. <a.s...@lo...>]

Thanks a lot !

	


    AureSÉGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org <//www.locongres.org>
address

	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/> 	



Le 04/04/2025 à 15:27, Tino Didriksen a écrit :
> I have added you both as Maintainer for apertium-fra, -oci, -cat, 
> -oci-cat, and -oci-fra.
>
> And to the Apertium-Romance team.
>
> That should cover most languages you need.
>
> -- Tino Didriksen
>
>
> On Fri, 4 Apr 2025 at 15:16, Aure Séguier <a.s...@lo...> wrote:
>
>     Hi,
>
>     Could you add my colleague Vincent Gleizes
>     (v.g...@lo...) to the mailing list ? He asked some
>     weeks ago and he already has begun to work on Apertium, so he
>     already has some questions.
>
>     I would also need him to be added to the oci and the oci-fra
>     repositories of Apertium on Git. Is that possible ?
>
>     I don't know who is administrator of these repositories, now that
>     Hector Alòs doesn't work on it anymore. If you need someone to
>     manage permissions on it, I could do it, as I'm the main
>     contributor of the occitan language.
>
>     We would also need, if that's possible, for both of us to be added
>     to oci-cat repository (and maybe the "cat" one some day, or at
>     least to be in contact with someone who can add words in it). Who
>     do I have to ask to do so ?
>
>     Thanks,
>
>     Regards,
>
>     -- 
>     	
>
>
>         AureSÉGUIER
>
>     Responsabla del pòle informatic
>
>     Congrès permanent de la lenga occitana
>
>     	
>     	
>     	
>     	
>     mobilePhone
>
>     	+33 (0)5 32 00 00 64 <tel:+33%20(0)5%2032%2000%2000%2064>
>     website
>
>     	www.locongres.org <//www.locongres.org>
>     address
>
>     	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau
>
>
>
>
>
>     	
>     facebook <https://www.facebook.com/lo.congres> 	
>     	twitter <https://twitter.com/locongres> 	
>     	linkedin
>     <https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/>
>     	
>     	instagram <https://www.instagram.com/locongres/> 	
>
>
>
>     _______________________________________________
>     Apertium-stuff mailing list
>     Ape...@li...
>     https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>

Re: [Apertium-stuff] Question about metaparadigms

[Aure S. <a.s...@lo...>]

Hi

Yes, it's something like that. Actually, I would like to simplificate 
the management of verbs which doesn't use the same paradigm in different 
varieties :

   <pardef n="parl/i__vblex">
     <e><p><l>i</l><r>ar<s n="vblex"/><s n="pri"/><s n="p1"/><s 
n="sg"/></r></p></e>
   </pardef>

<pardef n="cr/e[M]ar__vblex">
     <e alt="oci"><p><l>è</l><r>e</r></p><i><prm/></i><par 
n="parl/i__vblex"/></e>
     <e alt="oci@gascon"><p><l>è</l><r>e</r></p><i><prm/></i><par 
n="parl/i__vblex"/></e>
</pardef>

<pardef n="p/e[NS]ar__vblex">
     <e alt="oci"><i>e<prm/></i><par n="parl/i__vblex"/></e>
     <e alt="oci@gascon"><i>e<prm/></i><par n="parl/i__vblex"/></e>
</pardef>

   <pardef n="L-alt_G-pasalt">
     <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
     <e alt="oci@gascon"><par n="p/e[NS]ar__vblex" prm="prm"/></e>
   </pardef>

<e lm="acceptar"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>

In oci, it would generate "accèpti" (according to cr/e[M]ar__vblex 
paradigm) and in oci@gascon it would generate "accepti" (according to 
p/e[NS]ar__vblex paradigm).

For now, I have to define 2 forms :
<e lm="acceptar" alt="oci"><i>acc</i><par n="cr/e[M]ar__vblex" 
prm="pt"/></e>
<e lm="acceptar" alt="oci@gascon"><i>acc</i><par n="p/e[NS]ar__vblex" 
prm="pt"/></e>

When we will have 5 varieties (and maybe 9 one day), it will be 
impossible...

Thanks


Le 02/04/2025 à 11:07, Kevin Brubeck Unhammer a écrit :

>> I call a metaparadigm with :
>> <e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
>> Inside, I would like to call another metaparadigm and passing it the
>> "prm" argument :
>>    <pardef n="L-alt_G-pasalt">
>>      <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
>>    </pardef>
>> But it doesn't seem to work, I have an error :
>> Error (4685): Undefined paradigm 'cr/e[M]ar__vblex__prm'.
>>
>> Can I pass the prm parameter to another paradigm ?
> Do I understand you correctly that you would like metadix
>
>
>    <pardef n="cr/e[M]ar__vblex">
>      <e><p><l>cre<prm/>ar</l><r>cremar<s n="vblex"/></r></p></e>
>      <!-- I just made something up here -->
>    </pardef>
>    
>    <pardef n="L-alt_G-pasalt">
>      <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
>    </pardef>
>
> <e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
>
> to be compiled/transformed to something equivalent to
>
>    <pardef n="cr/e[M]ar__vblex_pt">
>      <e><p><l>creptar</l><r>cremar<s n="vblex"/></r></p></e>
>    </pardef>
> <e lm="acceptar" alt="oci"><i>acc</i><par n="cr/e[M]ar__vblex_pt"/></e>
>
>
> ?
>
> That does sound like it would be useful.
>
>
> As you found out, prm="prm" won't work for that; I'm not sure but I
> don't think the current XSLT scripts support it, so we would have to
> expand the scripts to be able to pass on the prm parameter to
> sub-paradigms.
>
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff



	


    AureSÉGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org <//www.locongres.org>
address

	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/> 	



Le 02/04/2025 à 11:07, Kevin Brubeck Unhammer a écrit :
>> I call a metaparadigm with :
>> <e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
>> Inside, I would like to call another metaparadigm and passing it the
>> "prm" argument :
>>    <pardef n="L-alt_G-pasalt">
>>      <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
>>    </pardef>
>> But it doesn't seem to work, I have an error :
>> Error (4685): Undefined paradigm 'cr/e[M]ar__vblex__prm'.
>>
>> Can I pass the prm parameter to another paradigm ?
> Do I understand you correctly that you would like metadix
>
>
>    <pardef n="cr/e[M]ar__vblex">
>      <e><p><l>cre<prm/>ar</l><r>cremar<s n="vblex"/></r></p></e>
>      <!-- I just made something up here -->
>    </pardef>
>    
>    <pardef n="L-alt_G-pasalt">
>      <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
>    </pardef>
>
> <e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
>
> to be compiled/transformed to something equivalent to
>
>    <pardef n="cr/e[M]ar__vblex_pt">
>      <e><p><l>creptar</l><r>cremar<s n="vblex"/></r></p></e>
>    </pardef>
> <e lm="acceptar" alt="oci"><i>acc</i><par n="cr/e[M]ar__vblex_pt"/></e>
>
>
> ?
>
> That does sound like it would be useful.
>
>
> As you found out, prm="prm" won't work for that; I'm not sure but I
> don't think the current XSLT scripts support it, so we would have to
> expand the scripts to be able to pass on the prm parameter to
> sub-paradigms.
>
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Re: [Apertium-stuff] Permissions on repositories and mailing list

[Tino D. <ma...@ti...>]

I have added you both as Maintainer for apertium-fra, -oci, -cat, -oci-cat,
and -oci-fra.

And to the Apertium-Romance team.

That should cover most languages you need.

-- Tino Didriksen


On Fri, 4 Apr 2025 at 15:16, Aure Séguier <a.s...@lo...> wrote:

> Hi,
>
> Could you add my colleague Vincent Gleizes (v.g...@lo...) to
> the mailing list ? He asked some weeks ago and he already has begun to work
> on Apertium, so he already has some questions.
>
> I would also need him to be added to the oci and the oci-fra repositories
> of Apertium on Git. Is that possible ?
>
> I don't know who is administrator of these repositories, now that Hector
> Alòs doesn't work on it anymore. If you need someone to manage permissions
> on it, I could do it, as I'm the main contributor of the occitan language.
>
> We would also need, if that's possible, for both of us to be added to
> oci-cat repository (and maybe the "cat" one some day, or at least to be in
> contact with someone who can add words in it). Who do I have to ask to do
> so ?
>
> Thanks,
>
> Regards,
> --
> Aure SÉGUIER
>
> Responsabla del pòle informatic
>
> Congrès permanent de la lenga occitana
>
>
>
> [image: mobilePhone] +33 (0)5 32 00 00 64 <+33%20(0)5%2032%2000%2000%2064>
> [image: website] www.locongres.org <//www.locongres.org>
> [image: address] La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau
>
>
>
>
> [image: facebook] <https://www.facebook.com/lo.congres>
>
> [image: twitter] <https://twitter.com/locongres>
>
> [image: linkedin]
> <https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/>
>
> [image: instagram] <https://www.instagram.com/locongres/>
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>

[Apertium-stuff] Permissions on repositories and mailing list

[Aure S. <a.s...@lo...>]

Hi,

Could you add my colleague Vincent Gleizes (v.g...@lo...) to 
the mailing list ? He asked some weeks ago and he already has begun to 
work on Apertium, so he already has some questions.

I would also need him to be added to the oci and the oci-fra 
repositories of Apertium on Git. Is that possible ?

I don't know who is administrator of these repositories, now that Hector 
Alòs doesn't work on it anymore. If you need someone to manage 
permissions on it, I could do it, as I'm the main contributor of the 
occitan language.

We would also need, if that's possible, for both of us to be added to 
oci-cat repository (and maybe the "cat" one some day, or at least to be 
in contact with someone who can add words in it). Who do I have to ask 
to do so ?

Thanks,

Regards,

-- 
	


    AureSÉGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org <//www.locongres.org>
address

	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/> 	

Re: [Apertium-stuff] Changes planned for the occitan language, the oci-cat and the oci-fra pairs

[Aure S. <a.s...@lo...>]

Hi,

Thanks for your encouragement. We take note of your offer of help, we 
will probably ask you for advices sometimes.

Regards,

-- 

	


    AuraSEGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org
address

	La Ciutat - Creem! , 5-7 carrèra|rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/> 	



Le 01/04/2025 à 23:14, Kevin Brubeck Unhammer a écrit :
>> This year, we are launching two new projects on Apertium :
>> - We will add two occitan dialects (Limousin and Provençal) to the 3
>>    that already exists (Languedocien, Gascon, Aranese) in the
>>    monolingual resources and for the occitan-french pair
>> - We will expand the monolingual dictionary for the Aranese dialect
>>    and expand the Occitan-Catalan pair (for Aranese and Languedocien
>>   dialects).
> That sounds great :-D
>
>> As I mentioned in a previous e-mail, if we further enrich the
>> monolingual dictionary as it is today, this will cause problems (slow
>> compilation, poorly readable data for humans who work on it). That's
>> why we're planning to change the way occitan dialects are managed by
>> Apertium. We're going to switch to a ‘feature’ system like the one
>> used for Catalan. We're going to define default features for each
>> dialect, so that the commands used until now continue to work. This
>> should not change anything for those who use the translator.
> Good to see more pairs using this, let me know if you need any help with
> the setup. We should probably look into a slightly more usable UI for
> the features too; at least the way they are currently presented on
> apertium.org gets a bit confusing with very many features.
>
>
> best regards,
> Kevin Brubeck Unhammer
>
>
>
> _______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Re: [Apertium-stuff] Question about metaparadigms

[Kevin B. U. <unh...@fs...>]

> I call a metaparadigm with :
> <e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
> Inside, I would like to call another metaparadigm and passing it the
> "prm" argument :
>   <pardef n="L-alt_G-pasalt">
>     <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
>   </pardef>
> But it doesn't seem to work, I have an error :
> Error (4685): Undefined paradigm 'cr/e[M]ar__vblex__prm'.
>
> Can I pass the prm parameter to another paradigm ?

Do I understand you correctly that you would like metadix


  <pardef n="cr/e[M]ar__vblex">
    <e><p><l>cre<prm/>ar</l><r>cremar<s n="vblex"/></r></p></e> 
    <!-- I just made something up here -->
  </pardef>
  
  <pardef n="L-alt_G-pasalt">
    <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
  </pardef>

<e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>

to be compiled/transformed to something equivalent to

  <pardef n="cr/e[M]ar__vblex_pt">
    <e><p><l>creptar</l><r>cremar<s n="vblex"/></r></p></e> 
  </pardef>
<e lm="acceptar" alt="oci"><i>acc</i><par n="cr/e[M]ar__vblex_pt"/></e>


?

That does sound like it would be useful.


As you found out, prm="prm" won't work for that; I'm not sure but I
don't think the current XSLT scripts support it, so we would have to
expand the scripts to be able to pass on the prm parameter to
sub-paradigms.

Re: [Apertium-stuff] Changes planned for the occitan language, the oci-cat and the oci-fra pairs

[Kevin B. U. <unh...@fs...>]

> This year, we are launching two new projects on Apertium :
> - We will add two occitan dialects (Limousin and Provençal) to the 3
>   that already exists (Languedocien, Gascon, Aranese) in the
>   monolingual resources and for the occitan-french pair
> - We will expand the monolingual dictionary for the Aranese dialect
>   and expand the Occitan-Catalan pair (for Aranese and Languedocien
>  dialects).

That sounds great :-D

> As I mentioned in a previous e-mail, if we further enrich the
> monolingual dictionary as it is today, this will cause problems (slow
> compilation, poorly readable data for humans who work on it). That's
> why we're planning to change the way occitan dialects are managed by
> Apertium. We're going to switch to a ‘feature’ system like the one
> used for Catalan. We're going to define default features for each
> dialect, so that the commands used until now continue to work. This
> should not change anything for those who use the translator.

Good to see more pairs using this, let me know if you need any help with
the setup. We should probably look into a slightly more usable UI for
the features too; at least the way they are currently presented on
apertium.org gets a bit confusing with very many features.


best regards,
Kevin Brubeck Unhammer 

[Apertium-stuff] Question about metaparadigms

[Aure S. <a.s...@lo...>]

Hi,

I have a question about metaparadigms, as I'm working in reorganize 
varieties in Occitan monodix.

I call a metaparadigm with :
<e lm="acceptar" alt="oci"><i>acc</i><par n="L-alt_G-pasalt" prm="pt"/></e>
Inside, I would like to call another metaparadigm and passing it the 
"prm" argument :
   <pardef n="L-alt_G-pasalt">
     <e alt="oci"><par n="cr/e[M]ar__vblex" prm="prm"/></e>
   </pardef>
But it doesn't seem to work, I have an error :
Error (4685): Undefined paradigm 'cr/e[M]ar__vblex__prm'.

Can I pass the prm parameter to another paradigm ?

Actually, I would like to create a metaparadigm which call different 
paradigms for each variety :

   <pardef n="L-alt_G-no-alt">
     <e alt="oci"><par n="cr/e[M]ar__vblex"/></e>
     <e alt="oci@gascon"><par n="p/e[NS]ar__vblex"/></e>
   </pardef>

   <pardef n="L-no-alt_G-alt">
     <e alt="oci"><par n="p/e[NS]ar__vblex"/></e>
     <e alt="oci@gascon"><par n="cr/e[M]ar__vblex"/></e>
   </pardef>

   <pardef n="L-alt_G-alt">
     <e alt="oci"><par n="cr/e[M]ar__vblex"/></e>
     <e alt="oci@gascon"><par n="cr/e[M]ar__vblex"/></e>
   </pardef>
That's why i tried the solution above, but maybe there is a better way 
to do this. I would take any advice on it.

Thanks a lot

Regards,

Aure Séguier

-- 
	


    AureSÉGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org <//www.locongres.org>
address

	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/> 	

Re: [Apertium-stuff] Are you still interested in reviewing/merging the i18n feature of apertium tools?

[Tino D. <ma...@ti...>]

Yes, absolutely still interested in getting this merged.

-- Tino Didriksen


On Wed, 12 Mar 2025 at 14:50, Ahmed Siam <aah...@gm...> wrote:

> Hi Tino,
>
> I intend to continue working on this feature in the following days
> (Hopefully getting it ready for review within a week).
>
> I wanted to make sure that you are still interested in it now.
> You sent me an email about 1 year ago that you will have a look
> at it when I say it is ready. Is this still applicable now?
> I would be happy If I could get it merged and it became useful for others.
>
> In case you don't remember it, this article may remind you:
>
> https://ahmedsiam.com/gsoc-2023-report-internationalization-of-apertium-tools
>
> Best Regards,
> Ahmed Siam
>

[Apertium-stuff] dev.apertium.org for Apertium dev/collab work

[Tino D. <ma...@ti...>]

Reposted now that dev.apertium.org is refreshed (see prior badblocks mail)
and almost entirely wiped. Machine has new IP (65.108.233.184,
2a01:4f9:1a:ad8f::2), but same SSH key.

If someone needs a powerful machine to run tests on or collaborate on other
Apertium-related tasks, there is one available over SSH at dev.apertium.org

Specs:
- AMD Ryzen 5 3600 @ 3.6 GHz (6c/12t)
- 64 GB RAM
- 2x 512 GB NVMe SSD in RAID0, of which 2x16 GB is swap
- OS: Ubuntu 24.04 LTS, with Docker installed

So far, these people have SSH root access via SSH keys (no passwords
involved):
- Tino Didriksen
- Nick Howell (nlhowell)
- Kevin Brubeck Unhammer (Unhammer)
- Jonathan North Washington (firespeaker)
- Sushain Cherivirala (sushain)
- Francis M. Tyers (spectie)
- Xavi Ivars (xavivars)
- Daniel Swanson (popcorndude / mr-martian)
- Tommi Pirinen (Flammie)
- Tanmai Khanna (khannatanmai)
- Vivek Vardhan Adepu (vivekvelda / naan_dhaan)
- Kartik Mistry (kart_)
- Eckhard Bick
- Lars Nygaard
- Trond Trosterud

Feel free to add more people. Just don't step on others' work, and try to
keep the .ssh/authorized_keys file organized. Add users and/or use Docker
to keep your tests contained so they don't infest the whole machine.

And be mindful this is a scratch space server. No backups, no permanent
sites, everything is subject to nukes.

-- Tino Didriksen

[Apertium-stuff] dev.apertium.org badblocks

[Tino D. <ma...@ti...>]

Both SSDs on dev.apertium.org are experiencing multiple bad blocks, so
I'll be wiping and refreshing that machine.

Bad blocks may have something to do with being run in mdadm RAID0
without trim/discard for years, leaving the SSDs with no free blocks
to ever reallocate into. I've certainly never seen this many and
sudden errors across multiple devices before.

After fstrim, first drive went from 8 to 3 bad blocks, and second from
25 to 12. But the damage is done - have to fully wipe, trim, test
again, and maybe reprovision.

Even though I've reminded everyone from the start that
dev.apertium.org is a scratch server and you can't trust any data to
stay on it, I'll take a backup of what I can, just in case.

-- Tino Didriksen

Re: [Apertium-stuff] ISSUE BUILDING APERTIUM-ENG-DEU

[Daniel S. <awe...@gm...>]

That's a warning, not an error. I see no errors in your log output. What
happens if you run a translation command like

echo "good morning" | apertium -d . eng-deu

Daniel

On Thu, Mar 13, 2025 at 6:01 AM NIRVANA <nirvana@userchef.love> wrote:

> HELLO, I AM ATTEMPTING TO BUILD THE LANGUAGE PAIR APERTIUM-ENG-DEU...
> I HAVE INSTALLED THE DEPENDENCIES...
> I BUILD LIKE THIS...
> git clone https://github.com/apertium/apertium-eng-deu
> cd apertium-eng-deu
> ./autogen.sh
> make
> MAKE FAILS WITH A FEW ERROR MESSAGES...HERE IS THE LAST LINE...
> "Debug mode name eng-deu-autoseq generated multiple times, disregarding
> result from eng-deu step 11"
> I HAVE ATTACHED THE LOG FILE...
>
>
>
> ~ U-SER-CHEF_______________________________________________
> Apertium-stuff mailing list
> Ape...@li...
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>

Re: [Apertium-stuff] Apertium Inquiry

[Tino D. <ma...@ti...>]

As https://github.com/apertium/apertium-apy/blob/master/apertium_apy/mode_search.py#L38
indicates, spamedical is not a valid identifier. Variants are _
suffixed, so it should be spa_medical and ast_medical

-- Tino Didriksen


On Thu, 13 Mar 2025 at 10:38, Víctor Sandín <vs...@as...> wrote:
>
> Thank you very much for your response.
>
> Checking what you say, I copied the modes directly to the other path and now apertium -l does get the modes including the one I created (previously it was *). But for example, if I do:
>
>
>
> curl http://localhost:2737/listPairs
>
>
>
> {"responseData": [{"sourceLanguage": "spa", "targetLanguage": "ast"}, {"sourceLanguage": "ast", "targetLanguage": "spa"}], "responseDetails": null, "responseStatus": 200}
>
>
>
> In the Dockerfile, I launch it with:
>
>
>
> ENTRYPOINT ["python3", "/root/apertium-apy/servlet.py", "--lang-names", "/root/apertium-apy/langNames.db"]
>
> CMD ["/usr/local/share/apertium/modes", "--port", "2737"]
>
>
>
> So I understand that it should use that modes path. I don't know if I should add something to the language_names folder (I tried without success).

Re: [Apertium-stuff] Apertium Inquiry

[Víctor S. <vs...@as...>]

Thank you very much for your response.

Checking what you say, I copied the modes directly to the other path and now apertium -l does get the modes including the one I created (previously it was *). But for example, if I do:

curl http://localhost:2737/listPairs

{"responseData": [{"sourceLanguage": "spa", "targetLanguage": "ast"}, {"sourceLanguage": "ast", "targetLanguage": "spa"}], "responseDetails": null, "responseStatus": 200}

In the Dockerfile, I launch it with:

ENTRYPOINT ["python3", "/root/apertium-apy/servlet.py", "--lang-names", "/root/apertium-apy/langNames.db"]
CMD ["/usr/local/share/apertium/modes", "--port", "2737"]

So I understand that it should use that modes path. I don't know if I should add something to the language_names folder (I tried without success).

De: Tino Didriksen <ma...@ti...>
Fecha: jueves, 13 de marzo de 2025, 9:51
Para: [apertium-stuff] <ape...@li...>
CC: Víctor Sandín <vs...@as...>
Asunto: Re: Apertium Inquiry
If you used the mailing list, more people would be able to answer.

APy from apt-get by default looks in /usr/share/apertium/modes. A pair
built from source and "make install" by default installs into
/usr/local and is thus not visible to /usr

You can run the pair's configure with --prefix=/usr to change where it
is installed (but this is strongly discouraged - only do that in a
container), or you can point APy's systemd service at /usr/local
instead (also container), or you can run APy manually and point it at
/usr/local

-- Tino Didriksen

On Thu, 13 Mar 2025 at 07:44, Víctor Sandín <vs...@as...> wrote:
>
> Hi again Tino,
>
>
>
> I'd like to ask a question, see if you can help. It turns out that if I want to generate a new dictionary with the language pair spamedical|astmedical (they use spa and ast for monolingual), I run the following command from the container terminal to Apertium:
>
> echo 'Cuestion medica' | apertium -d /usr/local/share/apertium spamedical-astmedical
>
> I do get the expected output for the dictionary I created for that pair, but if I make the request with curl, it tells me the pair isn't installed (if I do listPairs, it doesn't appear either).
>
>
>
> From what I've been checking, I've tried adding the lines:
>
> frp spamedical spamedical
>
> frp astmedical astmedical
>
> in the manual-additions.tsv file, but I haven't found a solution.
>
>
>
> If you could help me, it's for a university project, and I'm a bit desperate.
>
>
>
> Best regards and best wishes.

Re: [Apertium-stuff] Apertium Inquiry

[Tino D. <ma...@ti...>]

If you used the mailing list, more people would be able to answer.

APy from apt-get by default looks in /usr/share/apertium/modes. A pair
built from source and "make install" by default installs into
/usr/local and is thus not visible to /usr

You can run the pair's configure with --prefix=/usr to change where it
is installed (but this is strongly discouraged - only do that in a
container), or you can point APy's systemd service at /usr/local
instead (also container), or you can run APy manually and point it at
/usr/local

-- Tino Didriksen

On Thu, 13 Mar 2025 at 07:44, Víctor Sandín <vs...@as...> wrote:
>
> Hi again Tino,
>
>
>
> I'd like to ask a question, see if you can help. It turns out that if I want to generate a new dictionary with the language pair spamedical|astmedical (they use spa and ast for monolingual), I run the following command from the container terminal to Apertium:
>
> echo 'Cuestion medica' | apertium -d /usr/local/share/apertium spamedical-astmedical
>
> I do get the expected output for the dictionary I created for that pair, but if I make the request with curl, it tells me the pair isn't installed (if I do listPairs, it doesn't appear either).
>
>
>
> From what I've been checking, I've tried adding the lines:
>
> frp spamedical spamedical
>
> frp astmedical astmedical
>
> in the manual-additions.tsv file, but I haven't found a solution.
>
>
>
> If you could help me, it's for a university project, and I'm a bit desperate.
>
>
>
> Best regards and best wishes.

[Apertium-stuff] Are you still interested in reviewing/merging the i18n feature of apertium tools?

[Ahmed S. <aah...@gm...>]

Hi Tino,

I intend to continue working on this feature in the following days
(Hopefully getting it ready for review within a week).

I wanted to make sure that you are still interested in it now.
You sent me an email about 1 year ago that you will have a look
at it when I say it is ready. Is this still applicable now?
I would be happy If I could get it merged and it became useful for others.

In case you don't remember it, this article may remind you:
https://ahmedsiam.com/gsoc-2023-report-internationalization-of-apertium-tools

Best Regards,
Ahmed Siam

Re: [Apertium-stuff] Changes planned for the occitan language, the oci-cat and the oci-fra pairs

[Anthony J. B. <an...@an...>]

Aure Séguier writes:
> As I mentioned in a previous e-mail, if we further enrich the 
> monolingual dictionary as it is today, this will cause problems (slow 
> compilation, poorly readable data for humans who work on it). That's why 
> we're planning to change the way occitan dialects are managed by 
> Apertium. We're going to switch to a ‘feature’ system like the one used 
> for Catalan. We're going to define default features for each dialect, so 
> that the commands used until now continue to work. This should not 
> change anything for those who use the translator.

This will be a welcome change for OpenBSD. The excessive build times
of the occitan pairs have impacted our ability to provide them in our
package repositories.

[Apertium-stuff] Changes planned for the occitan language, the oci-cat and the oci-fra pairs

[Aure S. <a.s...@lo...>]

Hi,

For those who don't know me, I'm Aure Séguier. I work for Lo Congrès, 
which contributed a lot some years ago to the occitan language and to 
the occitan-french pair.

This year, we are launching two new projects on Apertium :
- We will add two occitan dialects (Limousin and Provençal) to the 3 
that already exists (Languedocien, Gascon, Aranese) in the monolingual 
resources and for the occitan-french pair
- We will expand the monolingual dictionary for the Aranese dialect and 
expand the Occitan-Catalan pair (for Aranese and Languedocien dialects).

As I mentioned in a previous e-mail, if we further enrich the 
monolingual dictionary as it is today, this will cause problems (slow 
compilation, poorly readable data for humans who work on it). That's why 
we're planning to change the way occitan dialects are managed by 
Apertium. We're going to switch to a ‘feature’ system like the one used 
for Catalan. We're going to define default features for each dialect, so 
that the commands used until now continue to work. This should not 
change anything for those who use the translator.

On the other hand, it will change the way data is managed on the 
backend. We'd like to know if there are people who would have a problem 
with this before we change everything. If so, please let us know.

If nobody objects, we'll set up the ‘features’ system. As well as being 
able to request a translation in a specific dialect, users will be able 
to add options so that the translation resembles their local sub-dialect 
as closely as possible (Occitan is a language with a great deal of 
intra-dialect variety).

I'll be working with a colleague of mine, Vincent Gleizes, who I'll let 
introduce himself. We are at your disposal if you would like more 
information on these upcoming projects.

Regards,

-- 
	


    AureSÉGUIER

Responsabla del pòle informatic

Congrès permanent de la lenga occitana

	

	
	

	
mobilePhone

	+33 (0)5 32 00 00 64 <tel:+33 (0)5 32 00 00 64>
website

	www.locongres.org <//www.locongres.org>
address

	La Ciutat - Creem! , 5-7 rue de la Fontaine, 64000 Pau





	
facebook <https://www.facebook.com/lo.congres> 	

	twitter <https://twitter.com/locongres> 	

	linkedin 
<https://www.linkedin.com/company/congres-permanent-de-la-lenga-occitane/> 	

	instagram <https://www.instagram.com/locongres/>