Re: Darwin: Replace environment runpath with embedded [PR88590]

["Nathanael Nerode" <ncn_gcc10@fastmail.fm>]

I hope a global maintainer can step up.  I've been on hiatus from GCC work for some years, and this was never my part of the build system anyway -- and I don't use Darwin -- so I'm not qualified to review it.  It looks fine but it should be reviewed by someone who knows what they're doing.

On Wed, Sep 20, 2023, at 9:52 AM, FX Coudert wrote:
> ping**2 for a build maintainer or global maintainer to review, please
> It restores testing on darwin >= 21, which is not currently working
>
> If no maintainer is available to review build system, could I get 
> guidance on how to proceed further?
>
> Thanks,
> FX
>
>
>> Le 12 sept. 2023 à 19:52, FX Coudert <fxcoudert@gmail.com> a écrit :
>> 
>> Hi build maintainers,
>> 
>> May I ping this series of patches for review? In particular, they allow to restore testing on darwin, which is currently broken with darwin >= 21, due to DYLD_LIBRARY_PATH being systematically removed from the environment by system tools.
>> 
>> It has been tested for two years on darwin, and would allow to restore regular regtesting on that target. It is a big step to help prevent bugs being undetected on this port.
>> 
>> The patchset was okayed from the driver point of view, but we need a build reviewer (or global reviewer) to okay those bits. Original presentation of the patches:
>> 
>> ------
>> I’d like to post an updated and rebased version of Iain Sandoe’s patches for modern darwin, originally posted in November 2021: https://gcc.gnu.org/pipermail/gcc-patches/2021-November/584775.html
>> 
>> The rationale in that message is pretty much unchanged, and the patches have been since tested thoroughly on darwin (both Intel and ARM) for almost two years now. We have been shipping Iain’s branch (including these patches) since then in Homebrew and most other major distros of GCC on Darwin. So I think it’s been very thoroughly tested.
>> 
>> The main comment that arose from review in the previous incarnation was the need to at least offer the libtool part of the patch to upstream, in order to reduce in the long term the divergence between our version and upstream. I have done so in https://savannah.gnu.org/patch/index.php?10385
>> 
>> (I would also note that I have offered other suggestions of small snippets that could be upstream in libtool for darwin, but have not received much feedback for now: https://savannah.gnu.org/patch/?10371)
>> ------
>> 
>> 
>> Thanks,
>> FX
>> 
>> 
>> 
>>> Le 29 août 2023 à 22:17, FX Coudert <fxcoudert@gmail.com> a écrit :
>>> 
>>>> I think a build machinery review is needed.
>>> 
>>> Thanks. CC’ing the relevant maintainers for review of the build part.
>>> The driver part and the darwin-specific part are already okayed.
>>> 
>>> FX
>
>
> Attachments:
> * 0001-Driver-Provide-a-spec-to-insert-rpaths-for-compiler-.patch
> * 0002-Darwin-Allow-for-configuring-Darwin-to-use-embedded-.patch
> * 0003-Darwin-rpaths-Add-with-darwin-extra-rpath.patch
> * 0004-Testsuite-allow-non-installed-testing-on-darwin.patch
> * 0005-Doc-document-the-new-Darwin-options.patch