On Sat, Mar 23, 2019 at 12:58:04AM +0100, Steffen Möller wrote: > > > I just commited my latest status. > Modded and pushed back.
Cool. > > The test suite is not running and > > my attempt to fix it does not work. Good luck with it. > On the command line it helped me to add the $(CURDIR)/.pybuild/... path > to the module to PYTHONPATH but I also did not end anywhere. It is > pretty close to uploadable, I think. The test, yes. I sense it to be a > dh_auto_test incompatibility that I am not prepared to investigate at > the moment. Instead, I am tempted to test it as part of a larger > workflow for now and upload. You can use it for the moment as local package. My impression is that ftpmaster is not as quick as before the freeze. I see chances that someone of our students might pick it up and I usually ask on debian-python list. Trying this and upload in 7-10 days if there is no solution woul be no harm. > >> I would otherwise fiddle with it here and there. > > Your ITP bug would have profited if you would just have run > > itp_from_debian_dir in the cloned dir. :-P > I was actually looking for it. Twice :) Where do you hide that beast? > Not even google knows. And neither does apt-cache search. $ git log | tail Author: Andreas Tille <[email protected]> Date: Fri Mar 22 09:28:38 2019 +0100 Initial packaging commit 4d90fbe076cc1e0b962891d8033940c070a46701 Author: Andreas Tille <[email protected]> Date: Fri Mar 22 08:16:51 2019 +0100 New upstream version 1.7 I was about to send an ITP. We had a real racing condition. > > PS: May be you can also have a look in to kallisto which just needs > > to get a sensible test. Snakemake input is not working - extracting > > the actual calls into a shell script could simplify things and does > > not add an extra dependency from kallisto. > > I was impressed by what you have already done for the package. I am not > available next week - I also suggest to commit it as it is if it is only > for the test that is problematic. We are in a freeze anyway and I am > more interested in seeing the different reverse dependencies finding > that package so we get some first workflows done. And test the package > with it, too. I decided to extract the single shell commands snakemake would trigger. There is no real point in creating an artificial snakemake dependency just to call a few command lines. The first one seems to be kallisto index -i transcripts.kidx transcripts.fasta.gz kallisto quant -i transcripts.kidx -b 30 -o kallisto_out --genomebam --gtf transcripts.gtf.gz --chromosomes chrom.txt reads_1.fastq.gz reads_2.fastq.gz which seems to be fine. However, before I messed up the command line and had kallisto quant -i transcripts.kidx -b 30 -o kallisto_out --genomebam --gtf transcripts.gtf.gz transcripts.fasta.gz --chromosomes reads_1.fastq.gz reads_2.fastq.gz [quant] fragment length distribution will be estimated from the data [index] k-mer length: 31 [index] number of targets: 14 [index] number of k-mers: 22,118 [index] number of equivalence classes: 20 Warning: could not find chromosomes for 117 transcripts [quant] running in paired-end mode [quant] will process pair 1: transcripts.fasta.gz reads_2.fastq.gz [quant] finding pseudoalignments for the reads ...Segmentation fault Segmentation fault is not good! I will check if I find some more tests and will report the latest issue upstream. Kind regards Andreas. -- http://fam-tille.de
