Hi Nadiya,

thanks for your work on this.  It seems you've hit a non-trival example
but that's no problem but rather a challenge, right?

At first I've fixed debian/changelog.  It seems you always called

    dch -i

which is for each call bumping the Debian release.  Please use the -i
option only once.  Even better use

    dch --team

since this adds the "Team upload" entry which marks the upload in a way
that lintian (the Debian policy checker) does not claim that the
changelog owner is not listed as Uploader in debian/control.

Please git pull to see the result.

On Thu, Mar 23, 2017 at 09:55:37PM -0700, Nadiya Sitdykova wrote:
> I've added tests of examples from here
> https://ginolhac.github.io/mapDamage/#a5 But only two of them works fine.

Hmm, I admit I get a strange error even here:

mapDamage -i rescale_test/pe_test/pe.sam -r rescale_test/pe_test/ref.fa
Started with the command: /usr/bin/mapDamage -i rescale_test/pe_test/pe.sam -r 
rescale_test/pe_test/ref.fa
        Reading from 'rescale_test/pe_test/pe.sam'
        Writing results to 'results_pe/'
pdf results_pe/Fragmisincorporation_plot.pdf generated
No length distributions are available, plotting length distribution only works 
for single-end reads
Error: Seqtk failed: [Errno 2] No such file or directory

I applied all patches via

   quilt push -a

and checked for occurences of seqtk via:

mapdamage(master) $ grep -Riw Seqtk | grep -v -e "^\." -e "^debian"
mapdamage/composition.py:    path_to_seqtk = 
mapdamage.rscript.construct_path("seqtk",folder="seqtk") 
mapdamage/composition.py:        sys.stderr.write("Error: Seqtk failed: %s\n" % 
(error,))
bin/mapDamage:    fpath_seqtk = '/usr/bin/seqtk'
bin/mapDamage:        sys.stderr.write("Seqtk executable not accessible; 
mapDamage has not\n"

So it seems that in mapdamage/composition.py the path to seqtk is not
properly assembled.  Would you mind checking this?

> Other two "Cannot use --<option> with missing R-libraries", so it seems
> they need recommended r-cran-rcppgsl package. I created simple stand-alone
> test with this package in depends to demonstrate that something goes wrong
> with installing RcppGSL. I looked at the source code here
> https://anonscm.debian.org/cgit/debian-med/mapdamage.git/tree/mapdamage/Rscripts/stats/checkLibraries.R
> and
> it looks like after package installation "library(RcppGSL)" doesn't load
> library correctly.

Hmmm, GSL rings a bell for me.  There was a gsl library transistion.
Are you using a current Debian testing or unstable system?  If not that
might be the cause of the issue.  May be you explain more verbosely
what exactly you mean by "doesn't load library correctly"?
 
I get

$ mapDamage --check-R-packages
All R packages are present


After checking mapdamage/rscript.py what R packages are really checked I
added some R packages as real Depends.  I personally follow the strategy
to use Depends if a package is needed to run the full test suite.  While
I actually learned from your commits that you can work around this via
"needs-recommends" (its nice to learn when mentoring ;-) ) I'm sure
users really want to have the full functionality.

In summary please do the following:

  1. Check why seqtk is not found in mapdamage/composition.py
     my guess is that the current quilt patch is incomplete.

  2. If the R test keeps on failing please provide more information

Thanks again for your work on this

     Andreas.

PS: Regarding the time delay you mentioned in your other mail:  That's
    fine - we have team members in different time zones. :-)

-- 
http://fam-tille.de

Reply via email to