guix-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug#55220] [PATCH 0/4] Add --list-systems and --list-targets options.


From: Mathieu Othacehe
Subject: [bug#55220] [PATCH 0/4] Add --list-systems and --list-targets options.
Date: Sat, 07 May 2022 18:04:26 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (gnu/linux)

> “as an argument to @option{--system}” maybe?

Done.

> This is my main issue: should we still accept any triplet, and simply
> print a nicer error than currently when the glibc dynamic linker name is
> unknown?
>
> Or should be be just as strict as above, at the risk of frustrating
> developers porting stuff to new or unusual platforms?
>
> Or should there be an option to bypass this check?

It's debatable for sure. However, now that glibc-dynamic-linker (and
system->linux-architecture in an additional patch) are relying on (gnu
platform) I think forcing the user to prompt a "supported" target, or to
define a new platform, makes sense.

If someone is adventurous enough to try something like: 'guix build
--target=new-target hello', patching glibc-dynamic-linker is already
most likely required. So now that glibc-dynamic-linker is relying on
(gnu platforms), defining a new platform will not be much harder.

> Maybe I’m overrating the usefulness of allowing users to pass in
> arbitrary triplets, though the manual does suggest that when porting to
> a new platform (info "(guix) Porting").  Thoughts?

I adapted this documentation section in v2.

> How about replacing:
>
>   (let ((s (false-if-exception (first (member arg (systems))))))
>     (if s …))
>
> with:
>
>   (if (member arg (systems)) …)

Done,

> Also, please remove final period in the error message passed to ‘leave’.

and done.

Thanks,

Mathieu





reply via email to

[Prev in Thread] Current Thread [Next in Thread]