upgpkg: python-{openstep-plist,uharfbuzz} diffenator2

[eclairevoyant]

Also looks like python-defcon in the repos is due for a python3.11 rebuild, as without the rebuild, python-ufo2ft fails to complete tests

[eclairevoyant]

Seems like diffenator2 only works with AUR/python-ninja and not [extra]/ninja for some reason, although I didn't really dig into it. I didn't replace it here in case I'm missing something obvious.

[alerque]

Thanks for these bumps. I will get them merged, but for a couple weeks I've been fighting with some issue in git subtree than is affecting aurpublish. An earlier merge commit of mine is corrupting the exported subtrees of all packages. I haven't wanted to reset them all and have a huge mess of duplicated commits here, but I also haven't found any way to backtrack or exclude the merge commit(s).

[alerque]

After weeks of frustration and searching I've given up trying to get git subtree to behave. It doesn't seem to have any mechanism for excluding/filtering a bad commit and not including it in the exported tree. I had two options: delete all packages from this repo and re-add them which would bring in 10+ duplicate commits, or force-push to rewind a few weeks of commits past the bad merge and re-import just the trees that had been touched since then. I opted for the latter route.

I can't merge this PR as is because it has the bad history, but I did cherry pick both of these bumps and they should be both in the AUR and in my prebuilt repository now. Thanks again.

[alerque]

I was able to build diffenator2 using [extra]/ninja without incident.

[eclairevoyant]

Building diffenator2 with either package works, I was referring to runtime. I can get the error if needed

[alerque]

Ninja is a build system, it souldn't be needed at runtime at all. If you are seeing an error lets get that filed upstream and/or on whatever package is relevant. If the ninja package is doing something wrong we can fix that too, but I suspect the error is actually on the upstream side at diffenator2 because it shouldn't require build system stuff at runtime.

[eclairevoyant]

Normally I'd agree, but it seems they've designed the program to use ninja at runtime.

rg -i ninja /usr/lib/python3.11/site-packages/diffenator2

Output:

``` /usr/lib/python3.11/site-packages/diffenator2/__main__.py 5:from diffenator2 import ninja_diff, ninja_proof, THRESHOLD, NINJA_BUILD_FILE 52: if os.path.exists(NINJA_BUILD_FILE): 53: os.remove(NINJA_BUILD_FILE) 54: if os.path.exists(".ninja_log"): 55: os.remove(".ninja_log") 58: ninja_proof( 71: ninja_diff( /usr/lib/python3.11/site-packages/diffenator2/__init__.py 4:from ninja.ninja_syntax import Writer 9:import ninja 17:NINJA_BUILD_FILE = "build.ninja" 20:def ninja_proof( 34: _ninja_proof(fonts, out, imgs, styles, filter_styles, characters, pt_size, user_wordlist) 44: _ninja_proof(fonts, o, imgs, styles, filter_styles, characters, pt_size, user_wordlist) 47:def _ninja_proof( 57: w = Writer(open(NINJA_BUILD_FILE, "w", encoding="utf8")) 89: ninja._program("ninja", []) 92:def ninja_diff( 110: _ninja_diff( 142: _ninja_diff( 157:def _ninja_diff( 171: w = Writer(open(NINJA_BUILD_FILE, "w", encoding="utf8")) 239: ninja._program("ninja", []) ```

Also, someone else filed an issue upstream (googlefonts/diffenator2#67) and a collaborator on the repo indicated that ninja was expected at runtime.

I'm thinking that AUR ninja package should be deleted as a duplicate and any fixes should be filed as a bug report?

[alerque]

I'm thinking that AUR ninja package should be deleted as a duplicate and any fixes should be filed as a bug report?

Yes, this is correct. I don't see any justification for it and duplicated packages like that are expressly forbidden in the AUR with specific exceptions which don't seem relevant there. If the [extra] package has a problem lets get a bug open on it and if needed I can even talk to the maintainer about it and see what we can do to fix it.