mpfi: fix download URL

[eduardosm]

Motivation for this change

The file_nr value was wrong, so it was downloading version 1.5.3 instead of 1.5.4.

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
• Built on platform(s)
  • NixOS
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Ensured that relevant documentation is up to date
• Fits CONTRIBUTING.md.

[eduardosm]

Result of nixpkgs-review pr 120614 run on x86_64-linux 1

2 packages failed to build:

• sage (sagemath)
• sageWithDoc

4 packages built:

• giac
• giac-with-xcas
• mpfi
• sollya

[7c6f434c]

@collares @omasanori @timokau any idea whether fixing the version bump is expected to break Sage? (or is it just an unrelated dependency failure?)

[eduardosm]

It looks like sage is currently failing to build on master too.

[eduardosm]

sage build logs:

• master: sage-master-log.txt
• this PR: sage-mpfi-log.txt

[collares]

Indeed, Sage is broken on master (probably due to the FLINT upgrade that happened a week ago). #116365 will unbreak it. Sage 9.3 ships with mpfi 1.5.2, but I don't think upgrading to 1.5.4 will be a big problem. If it is we can deal with it on the Sage side. Thanks for the heads-up!

[collares]

Wait, no, I am wrong: 4 tests fail on master, while a whole bunch of tests fail in the above log. If it's possible to wait a bit, landing #116365 first would save Sage a small headache.

[eduardosm]

I get lots of AttributeError: 'cypari2.gen.Gen' object has no attribute 'bnf_get_fu' when building locally, but I have no idea why. I'll run nixpkgs-review on #116365.

[collares]

Ah, I see, Sage's cypari2 was pinned to 2.1.1 to aid with the staging-next landing, but we migrated to 2.1.2 on master in the meantime and those changes conflicted with each other. Reverting commit c12ef05 should fix those bnf_get_fu errors.

@eduardosm If you do have a beefy machine, then a good way to ensure Sage is working is to revert c12ef05 and see if you only get 4 files with failing tests in Hydra (or testing on top of #116365). But if I had to bet, I would bet on the mpfi upgrade being safe, so if you don't have the resources to test Sage feel free to proceed with this PR.

[eduardosm]

I'll test on top of #116365 and report back.

[7c6f434c]

I bought the argument that «Sage on master is so broken, a bump to RC is an improvement». @collares it looks like you are currently the one with the best understanding of what is going on on the Sage side…

[collares]

@7c6f434c Thanks for merging! I opened #120648 for the cypari2 conflict (the problem started happening today with the staging-next merge), and after that Sage should be working on master again, so the tests for this PR should be reliable.

[eduardosm]

After applying #116365 and #120648, this PR will build fine.

[collares]

@eduardosm Many thanks for testing!

[timokau]

Thanks for fixing this @eduardosm. The nixpkgs-update bot makes the file_nr disclaimer easier to miss. We could opt-out for this package, but there is very little harm done either way.