[PATCH] D11737: Add -linker (and -linker=) alias for -fuse-ld=

[Richard Smith via cfe-commits]

On Tue, Aug 25, 2015 at 1:39 PM, Filipe Cabecinhas <
filcab+llvm.phabricator at gmail.com> wrote:

> Hi Richard,
>
> On Tue, Aug 25, 2015 at 11:01 AM, Richard Smith <richard at metafoo.co.uk>
> wrote:
>
>> On Aug 25, 2015 10:26 AM, "Filipe Cabecinhas" <
>> filcab+llvm.phabricator at gmail.com> wrote:
>> >
>> > Hi Richard,
>> >
>> > We wouldn't be able to link with libs matching
>> "libinker=*.{dylib,so,...}", I guess. If that is a big problem and you'd
>> prefer that we keep this as a private patch, let me know.
>>
>> I don't think it's a big problem, more just a "try to pick better flag
>> names in future" comment :) It sounds like you guys have existing systems
>> that depend on this name, so while I'm not really overjoyed about this,
>> accepting it for compatibility seems OK.
>>
> Thanks.
>
>> Can we produce an accompanying "deprecated" warning suggesting use of the
>> other name?
>>
> Not on our side. If you still want the warning, I'd prefer to just keep
> the flag private than to have the option+warning upstream (basically just
> for PS4 toolchain use) and have a private patch to remove the warning
> anyway.
>
> Let me know what you think,
>

What do you see as the future of this flag? If this is something you will
need to keep around essentially forever, then we should just take this
change into upstream clang. In that case, the patch LGTM.

If this is something you need for short-term compatibility until the
projects using it switch to -fuse-ld= or become obsolescent, then it makes
less sense to take it as an upstream change (if we accept it upstream
without a deprecated warning, it's likely that some other projects will
start to rely on it, which makes it harder to eventually remove).

  Filipe
>
> > Thank you,
>> >
>> >   Filipe
>> >
>> >
>> > On Mon, Aug 24, 2015 at 6:45 PM, Richard Smith <richard at metafoo.co.uk>
>> wrote:
>> >>
>> >> On Mon, Aug 24, 2015 at 5:50 PM, Eric Christopher via cfe-commits <
>> cfe-commits at lists.llvm.org> wrote:
>> >>>
>> >>> echristo added inline comments.
>> >>>
>> >>> ================
>> >>> Comment at: include/clang/Driver/Options.td:1853
>> >>> @@ -1853,1 +1852,3 @@
>> >>> +def fuse_ld_EQ : Joined<["-"], "fuse-ld=">, HelpText<"Use linker
>> <name>">, Group<f_Group>;
>> >>> +def linker_EQ : Joined<["-"], "linker=">, Alias<fuse_ld_EQ>,
>> MetaVarName<"<name>">;
>> >>>
>> >>> ----------------
>> >>> thakis wrote:
>> >>> > Any reason to have another alias for this at all? Does gcc
>> understand this spelling? If not, could you limit this flag to PS4 targets?
>> (I'm guessing you need it for PS4 targets for some reason.)
>> >>> Any reason? (And I'm not sure how to limit it btw on target).
>> >>
>> >>
>> >> -l already has a meaning; adding a different flag starting with -l is
>> a bad idea.
>> >
>> >
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20150825/bcb096ee/attachment-0001.html>