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

Re: Bug#618689: ITP: squizz bioinformatics sequence converter tool



Hi Andreas,
I am splitting squizz in packages squizz (binary), libsquizz and libsquizz-dev as you proposed.

I have an "issue" though, the generated libs (libsquizz) by the tool are libbioseq and libbioali, not some libsquizz-bioseq kind.

This raise a Warn in lintian.

Should I change the "make" files to modify the libs name, or is it acceptable anyway...

I could also create (sub)binary packages with package name libbioseq and libbioali, but I don't think it is acceptable to use several packages names (not like squizz-xxx). My control file would have source package squizz and binary packages squizz, libbioali, libbioseq

Thanks for your opinion

Olivier

Le 3/18/11 1:45 PM, Andreas Tille a écrit :
Hi Olivier,

thanks for working on this package.  (I actually have the impression
that Travemünde triggered some kind of competition who might package the
highest number of new packages. ;-))

I had a look into the packaging and I think it might work this way but
the existence of the libs leaves some space for enhancement.  In
addition I just commited some slight changes where I replaced general
comments in the boiler template - no need to talk about this further
here.

The problem is that if a software provides libraries these should be
provided as libfoo + libfoo-dev package while the former contains the
*.so.* files and the latter *.a + *.h files.  The binary tool which is
finally called by the user should go to either foo-tools.deb or in your
case simply squizz.deb.  So this is the general rule which you can read
in detail in the Library Packaging Guide[1].

There might be reasons to not follow this guide but these reasons should
be documented.  The fact that you included *.a and *.h files in the
package are in my eyes a clear sign that it might make sense that users
build their own programs linked against this library which makes perfect
sense to provide a libsquizz-dev package (or even a libbioali-dev +
libbioseq-dev ... whatever makes sense).

If those -dev packages would make sense then you should also have
separate library packages accordingly.  So the decision depends from the
fact whether such a development infrastructure should be provided or
not.  If it does not make sense the extra files should be left out of
the squizz packages (because we do not provide useless files).  If they
make sense we should (strongly) consider sticking to the suggested
package layout.

Kind regards and thanks for your work on squizz in any case

       Andreas.

On Fri, Mar 18, 2011 at 01:19:42PM +0100, Olivier Sallou wrote:
Hi,
squizz is packaged on SVN and has been tested in my config (and pbuilder
too).

Can someone check it and upload it, if ok ;-)
[1] http://www.netfort.gr.jp/~dancer/column/libpkg-guide/libpkg-guide.html


--
gpg key id: 4096R/326D8438  (pgp.mit.edu)
Key fingerprint = 5FB4 6F83 D3B9 5204 6335  D26D 78DC 68DB 326D 8438



Reply to: