[Bug jit/100613] libgccjit should produce dylib on macOS

["iains at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100613

Iain Sandoe <iains at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|dmalcolm at gcc dot gnu.org        |iains at gcc dot gnu.org
   Last reconfirmed|                            |2021-08-05
             Status|UNCONFIRMED                 |NEW
   Target Milestone|---                         |11.3
     Ever confirmed|0                           |1

--- Comment #2 from Iain Sandoe <iains at gcc dot gnu.org> ---
I have a preference for setting the correct library name when it is created,
and for setting the library versioning information in the 'usual' Darwin way.

The reason being that I have patches in flight for using "@rpath/....." for the
library names to fix the issues we have with testing on platforms with SIP
enabled (post 10.11).

Patches are under test.

NOTE there are issues with testing too - namely that the testsuite code is
currently making assumptions about support for --export-dynamic (it should
conditionally use -rdynamic IMO) and also that it assumes dejagnu.h is visible
in the default compiler lookup paths (it isn't on Darwin)... then we reach the
obstacle that the dejagnu.h from at least 1.6.2 has an unused variable in C
compilations + the wait function declaration clashes with the system's decl for
'wait'.

When all these things are fixed, we have two more issues that I've not been
able to address yet - namely that we get bad code-gen when "-g" is on the
command line, and that it seems that sometimes we get hanging tests that are
not killed by the testsuite machinery - so that the whole "make check" hangs.

At least I plan to address the issues for which the solution is clear - for the
last two things I think that more understanding of the processes used by the
Jit will be needed.