[PATCH wayland 1/4] wayland-util: do not export the wl_map_* API

[Daniel Stone]

Hi Emil,

On 30 August 2016 at 18:24, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> From: Emil Velikov <emil.velikov at collabora.com>
>
> Use only internally and explicitly marked as such with commit
> cf04b0a18f2 ("Move private definitions and prototypes to new
> zwayland-private.h")
>
> Signed-off-by: Emil Velikov <emil.velikov at collabora.com>
> ---
> I could not find any users of the API and I doubt there was ever one. If
> people feel nervous about this, we can keep it.

For the actual series, detaching the discussion from the
wayland-scanner bits in 0/4:

I think this one is fine, but I'd prefer to merge it at the same time
as the wayland-util split, if and when that happens.

Patch 2/4 (to move to the -private file) no longer applies, because we
let this series bitrot for so long. Sorry.

Patch 4/4 removes whitespace from the other -uninstalled.pc.in files,
but you add the same whitespace into the wayland-util file introduced
in 3/4 and don't fix it up in 4/4.

As for the actual libwayland-util split, I'm very much on the fence as
to whether it's a good idea. Broadly speaking I do like the idea and
sympathise with the aims, but am not sure how happy distro packagers
would be with an extra binary package to track. What really worries me
though, is transient symbol dependencies: at least with the pkg-config
modifications as-is in 3/4, with wayland-util dropping back to
Requires.private, I believe we'd see the compiler/linker complaining
that a project directly using symbols from wayland-util does not
directly link to it, only transiently via libwayland-{client,server}.
They could fix that by requiring wayland-util, but then they'd need
versioned fallbacks, and we've just made it a fair bit harder for
people to properly link to it.

Did you test with something that only has
wayland-client/wayland-server (and/or -uninstalled variants) in the
pkg-config file, directly using wl_list/wl_array/etc, and see if they
generated any warnings?

If we could be sure that was not the case, that would soothe my concerns a bit.

Cheers,
Daniel