gcc: model exported libraries, better mirrors, more specific license

[valgur]

Continues the changes from #15128 to model the installed runtime libraries correctly. The compilers cannot be used reliably unless self.requires("gcc/") with self.cpp_info.requires.append("gcc::gfortran") is used in addition to self.tool_requires("gcc/").

Set the package_type to library for the runtime libraries to be usable as self.requires(). See conan-io/conan#15306 for more context.

The current URLs for the sources were extremely slow - 10 minutes to download 140 MB. Replaced the .tar.gz with tar.xz for 80 MB instead of 140 MB size and picked the fastest 2 mirrors out of 25 tested. Download takes less than 10 seconds now.

Made the license ID more specific by including the exception for runtimes:
https://spdx.org/licenses/GCC-exception-3.1.html
https://gcc.gnu.org/onlinedocs/libstdc++/manual/license.html

Requires #21153 for Conan 2.0 pipeline to pass.

[github-actions]

🤖 Beep Boop! This pull request is making changes to 'recipes/gcc//'.

👋 @samuel-emrys you might be interested. 😉

[ghost]

I detected other pull requests that are modifying gcc/all recipe:

• [gcc] Model internal libraries #15128

This message is automatically generated by https://github.com/ericLemanissier/conan-center-conflicting-prs so don't hesitate to report issues/improvements there.

[samuel-emrys]

The changes you've pulled in from my PR in #15128 exceed the scope of what you've defined for this PR. Given that there seems to be some reluctance to adopt the internal library model for the gcc recipe, my suggestion would be to remove these changes from this PR if you want it to be accepted.

[valgur]

The changes you've pulled in from my PR in #15128 exceed the scope of what you've defined for this PR. Given that there seems to be some reluctance to adopt the internal library model for the gcc recipe, my suggestion would be to remove these changes from this PR if you want it to be accepted.

@samuel-emrys Yes, I actually care more about that change than the other minor ones, really.

gfortran is not really usable via a self.tool_requires("gcc/") unless self.requires("gcc/") with self.cpp_info.requires.append("gcc::gfortran") is also used. I've tried multiple approaches with self.cpp_info.system_libs and what not, and this is the only one that really works.

[conan-center-bot]

Conan v1 pipeline ❌

Failure in build 8 (f72ab0b9cee06e569070afdf27923c719de4c8dd):

• gcc/13.2.0:
  Error running command conan info gcc/13.2.0@#fa44be06c21a11c5a498218c294b8527 --json {jsonName} -pr {profileName}:

  [settings]
  arch=x86_64
  build_type=Release
  compiler=gcc
  compiler.libcxx=libstdc++11
  compiler.version=5
  os=Linux
  
  ...
  WARN: mpc/1.3.1: requirement mpfr/4.2.0 overridden by gcc/13.2.0 to mpfr/4.2.1 
  ERROR: Failed requirement 'isl/0.25' from 'gcc/13.2.0'
  ERROR: Unable to find 'isl/0.25' in remotes
  

• gcc/11.4.0:
  Didn't run or was cancelled before finishing

• gcc/12.3.0:
  Didn't run or was cancelled before finishing

• gcc/10.5.0:
  Didn't run or was cancelled before finishing

---

^{Note: To save resources, CI tries to finish as soon as an error is found. For this reason you might find that not all the references have been launched or not all the configurations for a given reference. Also, take into account that we cannot guarantee the order of execution as it depends on CI workload and workers availability.}

---

Conan v2 pipeline ❌

Note: Conan v2 builds are now mandatory. Please read our discussion about it.

The v2 pipeline failed. Please, review the errors and note this is required for pull requests to be merged. In case this recipe is still not ported to Conan 2.x, please, ping @conan-io/barbarians on the PR and we will help you.

See details:

Failure in build 8 (f72ab0b9cee06e569070afdf27923c719de4c8dd):

• gcc/13.2.0:
  Error running command conan graph info --requires gcc/13.2.0@#fa44be06c21a11c5a498218c294b8527 -f json -pr:h /home/conan/w/prod-v2/bsr/71378/defce/profile_linux_11_libstdcpp11_17_gcc_release_64..txt -pr:b /home/conan/w/prod-v2/bsr/71378/defce/profile_linux_11_libstdcpp11_17_gcc_release_64..txt:

  ======== Computing dependency graph ========
  ERROR: gcc/13.2.0: Package type is 'library', but no 'shared' option declared
  

• gcc/12.3.0:
  Didn't run or was cancelled before finishing

• gcc/11.4.0:
  Didn't run or was cancelled before finishing

• gcc/10.5.0:
  Didn't run or was cancelled before finishing

---

^{Note: To save resources, CI tries to finish as soon as an error is found. For this reason you might find that not all the references have been launched or not all the configurations for a given reference. Also, take into account that we cannot guarantee the order of execution as it depends on CI workload and workers availability.}

[jcar87]

Closing this, as per the reasons already expressed here: #15099 (comment)

Currently we don't have a package type that can correctly and comprehensively model libraries within a toolchain. In most circumstances, libraries within a toolchain like GCC don't need to be modelled: the compiler knows where to look for those libraries, and a "relocatable" gcc would know, as a fallback, where to look for libraries, or would already implicitly link them (e.g. like glibc, libstdc++, etc).

It is true that the gcc recipe in Conan center is unmaintained - at the moment we are not prioritising this because the vast majority of our users use a GCC provided by their system or SDK. If there are specific use cases that could be satisfied by an improved recipe, please open an issue where this can be discussed and other users can articulate their needs, and we will consider this in the future.

[valgur]

@jcar87 Thanks for looking into it. I agree with all of your points.

My real motivation in this PR was to get some kind of Fortran support going on CCI, which is required for quite a few numerical libraries. A full GCC toolchain with C/C++ support is more of a liability in that case anyway as it will interfere with the compiler configured in the profile. That's why I opened #23334. I invite you to take a closer look at it if you get a chance.

[jcar87]

Hi @valgur - could you please open an issue to discuss fortran, including which recipes would benefit for it, and on which platforms, and what would be the benefits to the wider user base? Thanks!