organizing the repo

[Udi-Fogiel]

Now that we have version tagging and date update, auto testing for regressions and building ctan archives using l3build, I think it will be a good time to clean the repository a bit.

Can we remove (or put aside) some of the unused scripts? Can we convert doc/TODO and doc/BUGS to tickets or regression tests? What about the misc directory, is it still needed?

[Udi-Fogiel]

In fact I believe the current situation is quite misleading. The .ldf files in the .tds.zip file are the one that are in the repo, and not the ones that generate from the .dtx file, although the .dtx file is in the source directory, so for example in MikTeX the .ldf files does not contain any of the license/copyright information, while if I will generate them from the "source" my self, they will.

Should they contain the comments that are add by the .dtx file?

[Udi-Fogiel]

Also the files generated from the .dtx file are all licensed under LPPL while the repo state polyglossia is licensed under MIT...

[Udi-Fogiel]

@jspitz do you know what ./tests/test-gloss-uyghur.tex tests? I'm not sure how to use it in the new test suit.

[jspitz]

@jspitz do you know what ./tests/test-gloss-uyghur.tex tests? I'm not sure how to use it in the new test suit.

AFAICS it tests some basic features with uyghur, including switch to LTR (english). Seems to be inspired from example-arabic.

BTW thanks for setting this suite up.

[Udi-Fogiel]

AFAICS it tests some basic features with uyghur, including switch to LTR (english). Seems to be inspired from example-arabic.

Then I'll probably move that file to the doc folder, with the rest of the examples.

BTW thanks for setting this suite up.

Sure, it was only fair after all the regressions I introduced :)

When you'll have time, can you look at doc/TODO and see what is still relevant, and maybe move whats relevant to tickets?

[jspitz]

I think TODO is all very old stuff. I would just remove it. Things that are still important today will be re-requested anyway.

[Udi-Fogiel]

what about gloss-occitan.dtx? are we using it?

[jspitz]

what about gloss-occitan.dtx? are we using it?

I think this is originally a separate package that has been included into polyglossia at some point (see also https://www.ctan.org/pkg/gloss-occitan).

I don't think we need this in our repository. @reutenauer?

Such external packages will break at some point with our rewrite (e.g., the keyval change). But I don't know why this package is not marked obsolete anyway.

[reutenauer]

I think this is originally a separate package that has been included into polyglossia at some point (see also https://www.ctan.org/pkg/gloss-occitan).

Something like that, yes – I can’t remember the details offhand but I think that the independent package came first, and I added it to the repository soon after that. I don’t think we need it either. I’ll try to answer the other questions arising out of this thread but I won’t have much time in the coming week.

[Udi-Fogiel]

I removed it at [a086aff]. What about branches? There are currently 37 of them, some does not seem the be ahead with any commits (probably merged), can I safely remove those?

[Udi-Fogiel]

What about all the lua module version? should we update it along the package version?

[Udi-Fogiel]

The remaining things are deciding about the lua module version, maybe port the scripts in ./tools to lua so that future development will have less dependencies (we will always have a lua interpreter...) and maybe moving uax9.sty elsewhere?

[jspitz]

What is the purpose of uax9.sty anyway?

[jspitz]

I mean, I understand what the package attempts do do, but this seems to be not released nor maintained, and it is also not included anywhere AFAICS.

[Udi-Fogiel]

The code looks rather similar to the one in https://github.com/khaledhosny/context-bidi. I've tried a couple of time to play with it, but even after making it work the result is not really adequate, maybe it is a good start though.

It would be nice to have a better support for bidi in LuaTeX, but maybe we can archive this file elsewhere. @reutenauer Do you have any opinion? I think you initially ported this code from conTeXt.