On Wed, 12 Nov 2025 02:08:05 +0100,
George Koehler <kernigh@gmail.com> wrote:
> 
> We patch meson to look at LIBname_VERSION (set by SHARED_LIBS in
> ports).  Our patch doesn't work when a meson.build has
> library(name_prefix: '').  Both net/dino and net/libupnpp have patches
> removing name_prefix: '' to work around this problem.  I offer this
> diff for name_prefix.
> 
> A typical library looks like
> 	library('what', sources)
> and prepends 'lib', so it is libwhat.so.${LIBwhat_VERSION}.
> 
> A library with named_prefix: might be
> 	library('libwhat', sources, name_prefix: '')
> which should have the same name, but might rename some generated
> files (like from 'what.pc' to 'libwhat.pc').  The version variable
> should be LIBwhat_VERSION, not LIBlibwhat_VERSION.
> 
> I rarely write Python, so it would help if someone checks my code.  It
> runs before the default self.prefix = 'lib' is set.  Given a typical
> library without name_prefix, hasattr(self, 'prefix') is false, so it
> should pick the same version variable as before.  The pick should
> change only if there is a name_prefix and the prefixed name begins
> with 'lib'.
> 
> This diff has built all the packages on my amd64 desktop, but has never
> been in a bulk.  I packaged net/libupnpp without its patch-meson_build.
> I tried to package net/dino without the 2nd chunk of
> patch-libdino_meson_build; it failed because some files changed names
> (dino.h to libdino.h and dino.vapi to libdino.vapi) and I had not
> updated PLIST.
> 
> Is this good enough to commit?

it allows to remove the hack from net/dino build. Which is good.

Code logical with one small remarks:

> ++                if prefixed_name[:3] == 'lib':
> ++                    libname = prefixed_name[3:]

I think that

+                libname = prefixed_name.removeprefix('lib')

is ideomatically cleaner.

with it, I see it worth to be commited because it allows to stop hacking
build systems. So, OK kirill@

-- 
wbr, Kirill