On 11/4/20 3:46 PM, Jakub Jelinek wrote:
On Wed, Nov 04, 2020 at 12:43:13PM -0800, Tom Stellard wrote:
>>> No, in that case gcc needs to Require: gas, because it is a run-time
>>> dependency of that package.
>>>
>>> CMake will still work if make is not installed. Packages that use
>>> cmake + Ninja should require those packages, and packages that use
>>> cmake + make should require those packages too.
>>
>> I don't choose either. I just run cmake and it chooses what to
>> use as it's default build tool. If I told it to use a specific
>> build tool then I would see your point, but for the default one
>> it seems reasonable that it should require it as it has made
>> the decision to use it.
>>
>
> Ok, I think this is a fair point especially after the f33 change were we are
> now using cmake --build and cmake --install instead of invoking make
> directly.
>
> The only issue is if cmake changes make to be a weak dependency can we still
> rely on this? Do brew builds automatically install weak dependencies?
>
> If the change was automated and you did not have to do anything would you
> still be opposed to having your spec files updated with BuildRequires: make
Well, gcc really should have either weak or strong dependency on make too
given that -flto is now used everywhere.
Thanks for clarifying. So it does sound like gcc will need at
dependency on make. If you do decide to use a weak dependency for this,
then I think I will need to update the proposal to BuildRequire: make
when gcc is used, so that we don't cause a performance regression in the
builds.
-Tom
Jakub