Hi Sascha and Anna,

at first thanks a lot to Sascha for answering close to all questions to
a level I would have had a hard time to cope with admittedly. :-)

I'm just picking those few items which are not answered yet.

On Wed, Mar 02, 2016 at 10:53:25PM +0000, Sascha Steinbiss wrote:
> 
> > 5. In the artfastqgenerator package I saw the following: "cp -a 
> > /usr/share/doc/artfastqgenerator/examples/*" Does it mean that these 
> > examples are provided by from upstream and they will be placed in that 
> > directory while package installation?
> 
> Thats’s what I expect, yes.

I confirm, yes.
 
> > 6. Tests for axe-demultiplexer change my font and background colors to blue 
> > and black respectively. Is something wrong with my system settings or is it 
> > an expected behaviour?

Strange.  For the moment I did the simplest way of testing by simply
calling the test script inside the Git repository:

axe-demultiplexer(master) $ debian/tests/gbs 
process_file -- (17:20:04) Starting demultiplexing

process_file -- (17:20:04) Finished demultiplexing

Run Summary:
Processed 0.00M read pairs in 0.0 seconds (175.8K read pairs/sec)
0.00M read pairs contained valid barcodes
0.00M read pairs could not be demultiplexed (100.0%)
[.] Destroying config structure



It looks a bit like "doing nothing" and I also can confirm that I get
lightblue text on black background.  Kevin, could you possibly comment
on this if this is the expected result?

Anna, if Kevin might skip this thread, I'd recommend to ask him as the
maintainer of the package who might most possibly know better than me.

> > 7. I ran test for eigensoft package (which checks that it prints usage 
> > info). It is marked as PASS, but stdout clearly says that it has an error: 
> > "error: eigenstrat do not print usage information". What's the matter? Why 
> > does it fail and why isn't this failure detected? (I guess the matter is in 
> > exit code -- it is alway equal to zero). Again, are my system settings 
> > incorrect or is it a common behaviour?
> 
> No idea about these, these seem to be quite package specific.

Eigensoft has no proper test suite provided by upstream - so this would
also be a target for you.  The "test" just checks that the executables
create "some" output - nothing else (not even matter whether the output
says anything about error).  This is not the best example for proper
testing but it at least verifies that the libraries the executables are
needing are installed and the executable throws no error code = better
than no test at all.

Regarding to the importance of eigensoft: Please put it close to the
bottom of your todo list - we have more important packages to test. :-)
 
> > 8. I'm looking for the test data for bwa. How do I check the licence for a 
> > dataset? May I use the test data from the other package? (Say bowtie2 -- it 
> > has test data inside). May I put the data in debian/tests or do I need to 
> > use some other location?
> 
> I wouldn’t put larger (say >1MB) test data into the debian/tests directory, 
> but rather create a bwa-test package containing the data and make it a 
> dependency of the test. Can anyone perhaps comment on this idea?

I frequently put the test data into an *-examples package.  The
rationale behind it is that the test script could also serve as an
example for the user.  The separation of binary and (larger) test
data is sensible, yes.

> IMHO you can use test data from other packages, but keep in mind that these 
> can change without warning.

+1
 
Thanks again to Sascha for his valuable comments

      Andreas.

-- 
http://fam-tille.de

Reply via email to