Skip to content

Conversation

raphael-proust
Copy link
Contributor

=== ERROR while compiling conf-gmp.1 =========================================#
 […]
== output ==
 + cc -c -I/usr/local/include test.c test.c: In function 'test': test.c:7:9: error: implicit declaration of function '__gmp_init'; did you mean '__gmpf_init'? [-Wimplicit-function-declaration]
     7 |         __gmp_init();
       |         ^~~~~~~~~~
       |         __gmpf_init

as seen in #28538

=== ERROR while compiling conf-gmp.1 =========================================#
 […]
== output ==
 + cc -c -I/usr/local/include test.c
 test.c: In function 'test':
 test.c:7:9: error: implicit declaration of function '__gmp_init'; did you mean '__gmpf_init'? [-Wimplicit-function-declaration]
     7 |         __gmp_init();
       |         ^~~~~~~~~~
       |         __gmpf_init
@jmid
Copy link
Member

jmid commented Sep 18, 2025

These past few days I am starting to see a number of failures caused by exploring conf-package lower bounds:

So it appears that something has changed, causing these to now surface and I am curious to understand what triggered it... 🤔

@jmid
Copy link
Member

jmid commented Sep 18, 2025

After writing the above I've created a dedicated issue for it in #28541 (as I originally intended) to help keep track.

@jmid jmid mentioned this pull request Sep 18, 2025
Copy link
Member

@jmid jmid left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK, after having dug into the rabbit hole of #28541 I agree with this change.

This should help avoid needless CI red flags.

@mseri mseri merged commit a8e705e into ocaml:master Sep 18, 2025
0 of 2 checks passed
@ejgallego
Copy link
Contributor

Hi folks, note that even after this PR, I could witness the failure in #28532 , unless I did something wrong in the rebase.

@jmid
Copy link
Member

jmid commented Sep 21, 2025

You are right, the avoid-version flag is not enough to handle this apparently. I also just saw it on #28560.

@raphael-proust I'm wondering what would be a good way to handle it then?

  • Change it to available: false instead?
  • retire the relevant conf- packages to https://github.com/ocaml/opam-repository-archive?
  • patch the lower-bound checks to respect avoid-version?
  • ...

@mseri
Copy link
Member

mseri commented Sep 22, 2025

I don't know if/how bad is to archive conf packages. As a first attempt, I'd aim at patching the lower-bound checks to respect avoid-version tags (unless it is absurdly complicated)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants