hdf5 pulls in gcc-wrapper

[shajra]

Issue description

The hdf5 library pulls in gcc-wrapper as a transitive dependency, which seems less than ideal.

The problem is that the build creates this text file summary of the build in lib/libhdf5.settings, that has the following:

Compiling Options:
------------------
               Compilation Mode: production
               C Compiler: /nix/store/kd7vb1afwvm2k65n9qjdw767w6lnwsrd-gcc-wrapper-6.4.0/bin/gcc ( gcc (GCC) 6.4.0)
               CFLAGS:
               ...

Additionally, this same text is embedded as a string in one of the shared libraries (lib/libhdf5.so.10.2.1).

The reason this is annoying for me is that I use transitive closures to bundle minimized tarballs, which I then put into Docker. But now gcc is grabbed, just because someone put in a informational string of which compiler was used.

I still have to go through the hdf5 source code and see if there's a simple fix. I could use the replace-literal utility to crop off the "/nix/store" from the reference in the shared library (maybe... not sure how tricky that might be). It's certainly easier to modify the libhdf5.settings file.

I'm open to other ideas.

Steps to reproduce

nix-store -qR "$(nix-build -E '(import <nixpkgs> {}).hdf5' --no-link)" | grep gcc-wrapper

yields

/nix/store/kd7vb1afwvm2k65n9qjdw767w6lnwsrd-gcc-wrapper-6.4.0

Technical details

• System: (NixOS: nixos-version, Ubuntu/Fedora: lsb_release -a, ...)

Debian testing with Nix

• Nix version: (run nix-env --version)

nix-env (Nix) 1.11.14

• Nixpkgs version: (run nix-instantiate --eval '<nixpkgs>' -A lib.nixpkgsVersion)

"18.03pre116475.d757d8142e"

• Sandboxing enabled: (run grep build-use-sandbox /etc/nix/nix.conf)

build-use-sandbox = true

[orivej]

Here removeReferencesTo is more appropriate than replace-literal.

[shajra]

@orivej cool, I just did that and it worked pretty well. I'll have a PR made up shortly and link to this ticket.

[shajra]

@FRidh thanks for taking care of the recommended changes (I was sleeping). I need to look up again the difference between native and normal build inputs. I still get confused about the difference.

[oxij]

A fun problem found via doCheckByDefault. With 82667fe applied h5cc fails with

h5cc: line 318: /nix/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-gcc-wrapper-7.3.0/bin/gcc: No such file or directory

in installcheck phase, and this eeeeeeeeeeeeeeeee* is made by remove-references-to.

I guess nobody uses binaries this package has in nixpkgs, but having those be broken by default isn't nice, I think.

I can't reopen this, but I suggest doing so and solving this by splitting the package into multiple outputs or something.

[stale]

Thank you for your contributions.

This has been automatically marked as stale because it has had no activity for 180 days.

If this is still important to you, we ask that you leave a comment below. Your comment can be as simple as "still important to me". This lets people see that at least one person still cares about this. Someone will have to do this at most twice a year if there is no other activity.

Here are suggestions that might help resolve this more quickly:

1. Search for maintainers and people that previously touched the related code and @ mention them in a comment.
2. Ask on the NixOS Discourse.
3. Ask on the #nixos channel on irc.freenode.net.

[Artturin]

No longer