Ludovic Courtès writes:

> Hi!
>
> Pierre Langlois <pierre.langlois@gmx.com> skribis:
>
>> Actually, thinking about this a little more, I'm not sure I understand
>> upstream decision to propagate -lz. The commit fixes [0] which indicates
>> it's so that taglib can be linked statically, but then that means if
>> we're dynamically linking, the application will also dynamically link
>> with zlib when it doesn't need to (at least not directly). And in guix
>> we only build shared libs for taglib so we're never statically linking
>> it AFAIK.
>>
>> So, here I'm a bit torn here, should we just follow what upstream is
>> indicating? Even it doesn't look right to me, but I might be wrong! Or,
>> should we revert the change that propagates -lz?
>
> I had the following patch that I intended to push, to avoid propagation.
>
> WDYT?
>
> Ludo’.
>
> commit d8124a707602980556fd33c7dbf9f7483fe1d0df
> Author: Ludovic Courtès <ludo@gnu.org>
> Date:   Mon Sep 7 09:56:08 2020 +0200
>
>     gnu: taglib: 'taglib-config --libs' shows -L flag for zlib.
>     
>     Fixes compilation of emacs-emms-print-metadata.
>     
>     * gnu/packages/mp3.scm (taglib)[arguments]: Add #:phases.

LGTM!

I was originally thinking we could just drop the `-lz`, since it
/should/ only be needed for people who statically link with taglib, and
we only ship shared libs. But actually, it's probably safer to follow
what upstream is doing.

Thanks,
Pierre