[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [GSoC] Check architectures



Hi Emmanouil,

On Fri, Jun 28, 2013 at 04:07:14PM +0300, Emmanouil Kiagias wrote:
> >
> > $ git diff
> > diff --git a/blend-gen-control b/blend-gen-control
> > index 1a81aa1..7e28ff0 100755
> > --- a/blend-gen-control
> > +++ b/blend-gen-control
> > @@ -86,7 +86,7 @@ class UDD_connector:
> >          """
> >          It queries the UDD and returns a list with all the available
> > Debian architectures
> >          """
> > -        query = "select distinct architecture from packages"
> > +        query = "select distinct architecture from packages where
> > architecture != 'all'"
> >
> >          self.__execute_query(query)
> >
> 
> Ok I made that change, thanks for your fast feedback :-)
> 
> I think I got the reason of the change, I will try to use proper wording:
> generating a control file for 'all' architecture(as I was doing previously)
> it will contain only the architecture independent packages which anyway we
> already include them in our query: blends-gsoc/sql/blendsd. And also in our
> case we want to create architecture dependent metapackages and independent.
> right?

I do not think that there is any sense in having separate metapackages
for Architecture=all.  The metapackages will be installed on a certain
architecture.  For this architecture (i.e. amd64) you have a metapackage
for instance

    med-bio_<version>_amd64.deb

there is no way to install any package

    med-bio_<version>_all.deb

in parallel.  So for every single architecture we need to create
metapackages featuring the packages for this architecture plus those
packages that are architecture = all which I expressed in my original
examle query in sql/blendsd_change_dependency by

  ...  architecture in ('$3', 'all' ))  ...

and which in your python code properly was turned into

  ...  architecture in ('{2}', 'all' )) ...

There is no point at all to turn this into

  architecture in ('all', 'all' ))

because there is no practical relevance for such a metapackage.

Kind regards

       Andreas.


-- 
http://fam-tille.de


Reply to: