ports.macports.org gives strange port dependencies

[Joshua Root]

On 2022-9-26 07:36 , Chris Jones wrote:
> 
> 
>> On 25 Sep 2022, at 10:27 pm, Joshua Root <jmr at macports.org> wrote:
>>
>> On 2022-9-25 01:13 , Christopher Jones wrote:
>>>>> On 24 Sep 2022, at 3:52 pm, Arjun Salyan <arjun at macports.org <mailto:arjun at macports.org>> wrote:
>>>>
>>>> The output of "port info pciids” from the same Docker container says:
>>> Can you please run
>>>> port -d info pciids
>>> pipe the output to file and send it here ? I want to see if we can figure out what is leading port to decide to add these deps.
>>
>> Compiler selection doesn't depend only on the platform, it looks at the actual versions (and existence) of the compilers available on the system. It's just another way in which a static PortIndex can't reflect the dynamic values of Portfile variables.
>>
>> So to make this work as you want, you'd have to either make a way to inject a fake compiler_version_cache, or provide compiler executables (or something that looks like them) that report the desired versions on the system running portindex.
> 
> Yeah I guessed as much.
> 
> Its always been a limitation of the site that it can only report the deps for a single platform, but its a bit of a shame that platform also isn’t actually a macOS version.
> 
> Instead of generating the portindex itself, running on linux, would it not work if instead it just grabbed the index from the macports buildbots, as also distributed to uses ? Say for the most recent OS. Using this instead of generating its own would at least make the results correct for one platform.

This should get rid of a lot of the questionable compiler deps on Linux 
at least: 
<https://github.com/macports/macports-base/commit/adb89e7e14d1ba6ac330077392eaef94e38f76cf>

- Josh