I realize that. But aren't files created in the /workspace also supposed to last past the current shell, so long as the image is around?

Thank you for reporting this! I'll look into it as part of this sprint towards v2.0 (haven't finished everything I wanted to include, so haven't announced it yet).

Fixed in Luthor v2.0! https://luthor.toolforge.org

Implemented in Luthor v2.0! https://luthor.toolforge.org

Implemented in Luthor v2.0! https://luthor.toolforge.org

Implemented in Luthor v2.0! https://luthor.toolforge.org

Implemented in Luthor v2.0! https://luthor.toolforge.org

Implemented in Luthor v2.0!

Implemented in Luthor v2.0!

Okay, I'm all set. For the benefit of anyone stumbling across this ticket: switching to envvars was simple enough, and the page I linked above gives the right recipe.

So, yeah, this page has all the clues.

Okay, it's coming back to me: I remember now that I need to use toolforge jobs. But trying to do so, and running bundle (Ruby equivalent of pip), it still fails due to missing *-dev packages. The first one is libyaml-dev. (for building the Psych gem, for pasing YAML.)

Looks like I'm all set. Thanks!

Okay, confirmed that it runs well on MariaDB as well, so no need for a VPS. Is there anything in particular I need to do to be able to host a 500MB database on MariaDB from my tool account?

Implemented but can't deploy yet due to T411226

Implemented but can't deploy yet due to T411226

Implemented but can't deploy yet due to T411226

Implemented but can't deploy yet due to T411226

Implemented but can't deploy yet due to T411226

Fixed but can't deploy yet due to T411226

Yes: the tool is a way to associate a usage example property with an existing lexeme in a given language. For the tool to be useful in the languages you mentioned, there first have to be some number of lexemes in those languages in Wikidata, that don't have a usage example already.

Hi, Agus! Do these languages have a Wikisource to draw usage examples from? If not, the tool would currently be useless.

Oh, then that sounds fine. The postgres requirement was the only reason I concluded I need a VPS project. The docs did not reveal the possibility of a Trove database for a Toolforge tool. That should work fine for me. And I'll also check whether the app can work with MariaDB without too much fuss, thanks. I'll update this ticket when I have news.

I think this is a very valuable proposal! The benefits are immense, and should certainly justify the nontrivial investment – it's clear this would require planning from SRE, async processing, UX design, etc., but, if implemented, it would add a simply vast amount of information to our discoverable offering, greatly increasing the use and value of our multimedia assets.

Me too. I still need to build Huggle myself every time, to get proper UTF-8 display on Linux (Mint). The snap binary does not display properly.

Thanks for the ping. We are indeed resolving it by using an address in learn.wiki. This ticket can be closed.

Can anyone provide an update on this?

Thank you! I confirm it now loads fine for me.

Loading the above translation still fails. Has this been deployed in production yet?

If we imagine a vocabulary-drill program, for instance, we would want it to know that the definite form of coronavirus is coronavirusset and not koronavirusset, i.e. to keep these sets of declension distinct. In other words, the lemma can be spelled either coronavirus or koronavirus, but if you pick one or the other, the rest of the declension must be consistent with it.

@Fnielsen – I think commingling alternate forms like that would hamper re-use, especially machine re-use: for instance, language acquisition tools.

Thank you, @Aklapper . In the comment to the patch, I see @ssastry saying, in May 2024, "Before we can merge this, we have to understand the implications of this change on Parsoid." Very well. In whose workstream is it to do that understanding? When can this patch be merged? Who needs pinging?

Implemented!

No need for an extension. The tool does not seem to be used much. I'll redeploy it on toolforge once we figure out why erex-yomi isn't working (see comment I tagged you on, on GitHub)