Re: gnupg2: fail with libgpg-error 1.47

[tlaronde%kergis.com@localhost]

On Thu, Jan 09, 2025 at 02:39:07PM +0100, Thomas Klausner wrote:
> On Thu, Jan 09, 2025 at 02:09:39PM +0100, tlaronde%kergis.com@localhost wrote:
> > I'm on:
> > 
> > NetBSD 10.99.12  x86_64
> > pkgsrc-2024Q4
> > 
> > FWIW, gnupg2 has:
> > 
> > BUILDLINK_API_DEPENDS.libgpg-error+=	libgpg-error>=1.46
> > 
> > But I had libgpg-error 1.47 and there was a build error with yat2m.
> > 
> > I upgraded to 1.51 (the version for pkgsrc-2024Q4), and the build
> > worked OK.
> > 
> > So it seems the version should be at least > 1.47. 1.51: OK. In
> > between: don't know.
> 
> I checked gnupg2's configure script and it claims it only needs 1.46:
> 
> NEED_GPGRT_VERSION=1.46
> 
> so this sounds like an upstream bug. Perhaps contact them for
> clarification which version is really needed?
> 
> Or look at the changelog/news for libgpg-error for the missing symbols
> (or whatever the exact error you see is).
> 

I will give it a look if I can find the time.

> > What is the policy regarding minimal versions for dependencies of a
> > package vs minimal version required by this package itself? If a
> > package depends on something that requires at least version Y, but,
> > by itself requires "only" version X, is the minimal required version
> > for the package set to the minimal version acceptable by the package
> > and all its dependencies?
> 
> When you build "something", it will pull in version Y, so when you
> later build "package" it will already have X installed.

OK, so let's look at the reverse: a dependency is happy with X and
build, then package depending on it requires Y, and when you replace X
by Y, you pull the carpet under dependency. How does it work in this
case?

-- 
        Thierry Laronde <tlaronde +AT+ kergis +dot+ com>
                     http://www.kergis.com/
                    http://kertex.kergis.com/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C