Re: Missing @INC path for pkg-perl-autopkgtest
On Sun, Oct 26, 2025 at 03:21:05PM +0100, gregor herrmann wrote:
> On Sat, 25 Oct 2025 15:03:22 +0200, Roland Rosenfeld wrote:
>
> > What do you think about this problem? Does it make sense to extend
> > pkg-perl-tools/autopkgtest to honor some debian/test/pkg-perl/set-env
> > variable? While writing this, I noted, that there already exists
> > debian/tests/pkg-perl/smoke-env for this, but this is only used in the
> > build-deps test.
>
> Ack, I think I've seen situations where I had set some env variable for the
> smoke test but couldn't do this for others.
>
> > Would you accept a merge request for pkg-perl-tools, that reads
> > debian/tests/pkg-perl/syntax-env in
> > runtime-deps-and-recommends.d/syntax.t like this?
>
> At first glance I see no problems with this approach. Let's see what others
> (like Niko, who wrote most of the autopkgtest stuff) have to say.
I wonder how much the pkg-perl-autopkgtest scripts should bend to cater
for other use cases than standard CPAN modules.
But given the current implementation is pretty much the antithesis of
clean and strict design principles, I don't mind really. Whatever works :)
That said, you (Roland) might want to consider /usr/share/chordpro/lib
or something like that. I don't see much point in installing modules
under the normal "public" search path while making them undiscoverable
with an extra subdirectory. Having them under /usr/share/chordpro would
make it clearer that they are private.
> Not sure if we want a specific debian/tests/pkg-perl/syntax-env or a generic
> debian/test/pkg-perl/{set-,}env for all tests …
I think a separate file for each test helps keep track of what's needed
for what. It's not like that's going to lead to much duplication anyway.
Obviously consistent naming would be nice, as far as possible. So I
guess debian/tests/pkg-perl/syntax-env would be my preference given the
existing names.
But it's not like I own this stuff, feel free to ignore me in case you
disagree :)
Happy hacking,
--
Niko
Reply to: