[new release] goblint-cil (2.0.4)

[sim642]

A front-end for the C programming language that facilitates program analysis and transformation

• Project page: https://github.com/goblint/cil

CHANGES:

• Add Return statement expression location (Location fixes for Goblint YAML witness generation/validation goblint/cil#167).
• Add availability attribute support (Everything fails on MacOS 14 because of attributes with assignment in time.h goblint/cil#168, Add assignment and floating-point constant attributes goblint/cil#171).
• Remove unused Libmaincil module (Remove unused Libmaincil module goblint/cil#165).
• Fix some synthetic locations (Fix first in location synthesizing goblint/cil#166, Location fixes for Goblint YAML witness generation/validation goblint/cil#167).

[sim642]

The freebsd (experimental) failures seem to be the same as last time: #24812 (comment).

For the goblint packages whose goblint-cil upper bounds this PR changes the failures are:

1. For freebsd, they're exception Failure("No good preprocessor (cpp) found"), somehow indicating that despite the conf-gcc dependency, there's no cpp binary around. Maybe FreeBSD's gcc package does not install that?
2. For macos, they're exception Goblint_lib__Maingoblint.FrontendError("Frontc.ParseError: Parse error") which this version of goblint-cil fixes (among other incompatible thing). These failures appear now because MacOS headers have been updated since.

[shonfeder]

Thanks for the error, analysis and the update, @sim642 :)

Assuming that these are not fundamental incompatibilities with the platforms failing on CI, do you think we should either the package to accept those CI failures in the future, and/or add a post-failure message on them indicating how a user could adjust their system to let the installation go thru?

[sim642]

I added some x-ci-accept-failures and availability changes that should be appropriate for these packages.

[mseri]

Thanks