CMake dependency data now available
Aleix Pol
aleixpol at kde.org
Tue Oct 14 14:19:33 BST 2014
On Sun, Oct 12, 2014 at 10:57 AM, Rolf Eike Beer <kde at opensource.sf-tec.de>
wrote:
> Am Sonntag 12 Oktober 2014, 21:24:51 schrieb Ben Cooksley:
> > Hi all,
> >
> > Not too long ago, sysadmin received a request to have the tool
> > "cmake-dependencies.py" from kde-dev-tools run by the CI system to
> > make packagers jobs easier.
>
> […]
>
> > If anyone has any queries, please let me know.
>
> This is a flat view on the dependencies AFAICT. In other words, you can't
> tell
> if this is a direct dependency or a "dependency [of a dependency]+", so
> this
> should not be used directly for a Depends: line IMHO. I was just looking at
> KGpg (what else *g*), and there show up things like Phonon, Cygwin and Perl
> that KGpg does not care in any way about, these things are just dragged in
> by
> something else.
>
> It's no problem that these things show up if you use this information as
> "what
> do I need to install to be sure that KGpg builds", but please don't use
> this
> as "these are the things that KGpg must directly depend on when installed".
>
Well, the fact that you as a KGpg maintainer don't care about, doesn't mean
that it's false.
Furthermore, this also displays what's the environment used while testing a
said package. KDE CI can't ensure things will work the same way if you have
different dependencies.
If you deem necessary, we can introduce a depth field. In fact, let me look
into it...
> Btw, is this line for Perl correct?
>
> {"files": ["FindPerl.cmake"], "version": "5.16.2()", "project":
> "Perl"},
>
> There are both "null" and "()" as version information for some packages, I
> guess they both mean "no specific version requested"? Should those be
> merged
> into the same value?
>
Yes. The parenthesis displays what part of the version is not required
IIRC, correct.
Aleix
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/distributions/attachments/20141014/3301c981/attachment.htm>
More information about the Distributions
mailing list