Sunday, July 07, 2024

Re: Update from maintainer: devel/objfw 1.1.5

On Sun, Jul 07, 2024 at 10:52:51PM +0200, Jonathan Schleifer wrote:
>
> > Am 07.07.2024 um 21:39 schrieb Theo Buehler <tb@theobuehler.org>:
> >
> > On Sun, Jul 07, 2024 at 12:33:38PM +0200, Jonathan Schleifer wrote:
> >>> Am 07.07.24 um 11:13 schrieb Theo Buehler:
> >>> Thanks. I see lots of changes in exported symbols that should usually
> >>> result in a major bump for libobjfw and libobjfwrt.
> >>
> >> Thanks for checking!
> >>
> >>> Is there some reason these bumps were left out?
> >>
> >> Yes, these were private symbols that were accidentally visible. Nothing
> >> should have used them.
> >
> > Thanks for the explanations. I don't think intention is good enough to
> > avoid a bump. These are still symbols that were exported and that could
> > (in principle) have been used even if they are in a private header that
> > wasn't installed. I could agree if all the symbols had been in the
> > private namespace, but that's not the case here.
>
> There is no private namespace in ObjC, unfortunately, as it doesn't have any namespaces just like C. The only way to avoid symbols being exported is to add __attribute__((visibility(hidden))) to every declaration of a private function, which is exactly what was added here.

I see.

> > So unless there is a really strong reason not to, I would suggest we
> > bump the major of both libraries.
>
> I'm pretty sure there was no usage of these in the wild, but I'm ok with a bump. Can you just do it on import or should I send another patch?

Let's just play it safe. I added the bump myself.

No comments:

Post a Comment