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

Re: HTSeq released incomplete tarball.

Hi Diane,

On Tue, Apr 08, 2014 at 11:22:56AM -0700, Diane Trout wrote:
> Hello,
> I was wondering what one should do when upstream's source release tarball is 
> missing important components.
> HTSeq's release tarballs after my initial release stopped including the .pyx 
> and .i source files and only included the C/CXX code generated by SWIG and 
> Cython.
> I also discovered its not longer including the documentation tree.
> Upstream pointed me at their subversion repository, but it doesn't even have 
> release tags.

:-(  Some upstream do not really evaluate our work to do us a favour to
do sensible release management.  At least some tags would be really a
good idea to not end up with some testing stuff they might have recently
injected.  My first action would be to ask them again for at least
tagging their releases.

> I've tried to help upstream by trying to make some suggestions 
> to improve their packaging. But I'm not sure what I should do now.
> My first attempt was to make a quilt patch that included the missing .i and 
> .pyx files but then I discovered that the doc tree was missing and I'm not 
> sure I want to add in a patch that is half of their release. (Upstream 
> tarballs dropped from 52k to 22k because of the missing docs and source 
> files.)

I admit that having patches larger than the original source does not
make any sense.
> I could make my own tarball from their subversion archive.
> Or I could keep trying to engage upstream to make a new release with the 
> missing files.

Both options are possible.
> Do you have any suggestions?

I'd suggest to do all discussion on the mailing list since

   a) we are informed what happens
   b) others might bring in helpful arguments

Without beeing able to evaluate your chances to convince upstream (which
would my *personal* prefered choice) I think you shoudl decide what from
your point of view is the solution that might create the least trouble
with an acceptable result for the user.  We have some
debian/get-orig-source scripts in SVN which could serve as an example.

Hope this helps



Reply to: