[Debian-med-packaging] Bug#846671: Bug#846671: Artemis should adapt to new htsjdk API which has dropped SAMFileReader (Was: [samtools/htsjdk] SAMFileReader vanished in Version 2.7.0 (#767))]

Thu Dec 8 08:11:05 UTC 2016

Ahh, I've found branch 6_0_17_fix - I can do the merge ...

On Thu, Dec 08, 2016 at 09:06:52AM +0100, Andreas Tille wrote:
> Hi Sascha,
> 
> On Thu, Dec 08, 2016 at 01:04:34AM +0100, Sascha Steinbiss wrote:
> > to address this problem I have taken a shot at patching Debian’s Artemis to use the new htsjdk API, avoiding SAMFileReader and using the SamReaderFactory instead. This fixed the FTBFS for me.
> > I tested BAM file reading by opening MAL1.embl.gz from the test/data directory and adding MAL1_8h.bam via ‘File->Read BAM/VCF...'. One of the genes has some mapped reads that are indeed shown. Comparing the displayed pile of mapped reads to the one shown by the recent Artemis version I have on Mac OS X, the result seems to be correct, but given my lack of practical experience with the BAM/VCF/‘anything-to-do-with-reads' components of Artemis I can’t say if I caught everything.
> 
> Thanks for this.  That's probably very helpful for upstream as well as
> the Debian package.
>  
> > I also updated the Debian version to 16.0.17, the latest release from Sanger. This allowed me to drop a couple of patches that I already merged earlier with my part-time-upstream hat on.
> > For now I have pushed my work into the ’6_0_17’ branch in git and I would like to kindly ask for some more testing. I don’t have suitable test data here and don’t really feel like an expert to test the right usage patterns. 
> 
> I do not find branch 6_0_17 and I do not even think that we need this
> extra branch.  I'd recommend to use master and as far as I understood
> Olivier's comment your test should be sufficient.  I personally also do
> not feel able to test but I think under this circumstances its a
> sensible approach to upload and thus enable some wider testing rather
> than expecting people to build a separate branch.
> 
> I have unmerged the fastqc and artemis bug since it seems we will be
> able to fix both packages without reintroducing the old API to htsjdk.
> 
> So I'd recommend you merge your separate branch back to master and push
> these changes.  I can have another look (I'm also currently bumping
> debhelper to compat level 10 and mark those watch files I have verified
> to version=4 just to have a marker even if version=3 works as well). I'm
> perfectly fine if you upload yourself without my additional inspection.
> 
> Thanks again for your very helpful contribution
> 
>      Andreas.
> 
> -- 
> http://fam-tille.de

-- 
http://fam-tille.de