Re: About ports QA by 'port test'

[email protected] Fri, 19 Mar 2010 23:07:30 +0200
Newsgroups gmane.os.freebsd.devel.quality-assurance
Message-ID <op.u9t2ashnmhpy7y@vova-vaio>
On Wed, 17 Mar 2010 14:58:49 +0200, Alexander Churanov <alexanderchuranov=
@gmail.com> wrote:

> Folks,
>
> If the portlint would be told to ignore errors like missing PORTVERSION=
,
> then how do you know whether the port is correct or not?
>
> From my point of view, the issue is the portlint pretends to parse
> makefiles, but does not implement full make :-). For example, as far as=
 I
> understand, the line 2188 just verifies that the file has a line starti=
ng
> with PORTVERSION. Portlint should use "make -V PORTVERSION" instead.

It's true that portlint parses makefiles. But I believe its task is check=
ing
style, not implementing make. Correctness is checked by the next stage of
the test. I have modified the script so that a FATAL error of portlint do=
es
not stop the test. One more file is used for results of portlint now.

New results will come to the site as the test is run over the list of
the installed ports.

> Alexander Churanov,
> maintainer of devel/boost-*

devel/boost* have style of implementing related ports really different
  from recommended in FreeBSD Porter's Handbook section 5.8 MASTERDIR.
So, the reason for complains from portlint is stylistic disagreement ;-)

--=20
Vladimir Chukharev
Tampere University of Technology
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-qa
To unsubscribe, send any mail to "[email protected]"