Re: How to deal with non-functional package in stretch? (Was: Bug#897605: profphd failure in stretch)

Hi All,

This is fixed in version 1.0.42-2, and not present in version 1.0.42-1, which seems to be latest in stretch (both with tests).

At patches dir:

https://salsa.debian.org/med-team/profphd/blob/master/debian/patches/2_remove_perlbrew_usage.patch

- this disables perlbrew usage (this was temporary, it would be wiser to replace this patch with the one which removes if-then-statement completely, because this "temporary" fix just confuses and intricates code logic in long-term perspective),

https://salsa.debian.org/med-team/profphd/blob/master/debian/patches/3_fix_indexing.patch - and as far as I remember this fixes actual erratic behavior with latest versions of perl, which was the reason why that if-then-else statement appeared in profphd code (the main problem here is the "$["-magic in perl).

Other patches fix error or warning messages, which appeared in almost all rostlab packages and were not as tricky as this one.

P.S. - I could work on fixing this bug late at the evening or tomorrow, because at this moment I have no Debian distribution or VM somewhere nearby, and I could check that everything works after I come home. In case if Liubov has time for solving this before that and she is not working on something else I've added her to CC.

2018-05-04 10:27 GMT+03:00 Andreas Tille <andreas@an3as.eu>:

Hi release team,

it turned out that profphd is non-functional in stretch (see bug
#897605). Since its not obviously a security issue we are wondering
whether there is some way to integrate this admittedly low popcon
package into Debian 9.5 (whenever this might be released).

While we can certainly do a backport not all users will realise / profit
from this.

On Fri, May 04, 2018 at 04:01:48AM +0000, olivier sallou wrote:
> Le jeu. 3 mai 2018 21:55, Andreas Tille <andreas@an3as.eu> a écrit :
>
> > On Thu, May 03, 2018 at 02:58:58PM +0000, olivier sallou wrote:
> > >
> > > newer version in sid (1.0.42-2) added a patch to allow new perl release.
> > A
> > > quick test on sid works (at least no error, just calling program with no
> > > argument), so maybe backporting this release would do the job
> >
> > Ahhh, right - I remember that Tatiana had to do some heavy patching.
> >
> > > > We definitely need some CI test for this package obviously.
> >
> > That was also done by Tatiana.
> >
> > However, a backport does not really close a bug in stretch and the issue
> > is also not really a security issue, right?
> >
>
> This is indeed not a security issue, but package is unusable. Backport
> would at least make software available.
> I don't know know what is usual debian policy regarding this use case. I
> suppose it occured multiple times.
>
> I can't be sure that sid version is really working, it just does not show
> any error. Could tatiana do some 'real' testing?

As far as I understood Tatiana she did some effort to make the program
functional. She has done some real life tests.

Kind regards

Andreas.

--
http://fam-tille.de

Best wishes,

Tanya.