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

Re: Packaging EMBASSY.

On Wed, 26 Sep 2007, David Paleino wrote:

2) change the debian package name, patching the software where needed, but this
is a really big effort.

Which I would only undergo if there is actually a name clash.  There is no
need to invent patches and problems before any real problem raises.  Educating
upstream is the more sane solution.  It might be that upstream agrees to
the strategy but is lacking man power and than we could propose patches but
should make sure that they will really accepted first hand.

I don't really like CDBS, it's like a black box to me. I mean, you just include
makefiles, and don't know what they're doing behind the curtains. I prefer
debhelper makefiles (the "classic" debian/rules with dh_* calls), which let you
define each action of the build process.

I have often heard this opinion but fail to fully understand the reasoning.
CDBS just does the same debhelper stuff - just avoids to repeat code in
a lot of debian/rules files.  So there is no hidden thing behind the
curtains - just read the included Makefiles.  The only problem are
mostly caused by complex packages that need several "non standard" stuff
and you will be forced to fiddle around with setting variables and
redefine targets.  It's sometimes a little bit hard to find out which
variables to set and which target is the correct one to redefine. In this case nobody will prevent you from not using CDBS and use plain
debhelper.  Thus I prefer CDBS if it fits to a simple and straigtforeward
package and debhelper if several tweaks are necessary because you tend
to loose the advantages of CDBS.

Calling CDBS magic behind the curtains would apply as well to the
debhelper scripts: Why not using plain shell programming instead of
using those magical dh_* scripts?

We should try, but after the first release we should contact upstream anyway,
there's no point in patching every single EMBASSY release to use plplot instead
of eplplot, if upstream doesn't wish so.


I believe that a release could be a good thing. Consider that, instead of just
us Debian-Med developers, we'll have a major public to test EMBOSS with. Most
than probably, there will be bugs filed which we didn't / don't notice, and
that's a good way to improve software. Well, at least IMO :)

Fully ACK.

Kind regards



Reply to: