[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: qiime(2) had unsatisfiable dependencies in d/control



Hi Liubov,

On Tue, Mar 12, 2019 at 12:47:59AM +0200, Liubov Chuprikova wrote:
> The old qiime(1) dependencies were not deleted from the d/control. I have
> pushed the required changes.

Thanks a lot for cleaning up behind me since I obviously did not my duty
as developer and installed the package before uploading it.  Since the
qiime package does not even install I filed bug #924373 (which is
actually fixed by your commit).  My bad for not checking earlier - so
qiime did not make it into testing which is really a shame.

This is another very good example that we are on one hand doing bad in
caring for our users sufficiently and on the other hand users care to
less to help us by simply testing what we are providing.

But back to the topic: You removed not only the dependencies that are
preventing qiime from installing successfully but also a lot of others.
I wonder whether Qiime2 is now independent from the lot of tools it has
used before.  Is there any relevant documentation or can we sneak into
the conda rules which might have the relevant metadata as well?  For
instance I cloned the bioconda repository on Github[1] but it only
contains rules for qiime version 1.9.1[2].  So my question is:  If
Qiime2 is propagating the installation via Conda is there some other
repository containing the conda metadata?  I could imagine that we can
learn from this about how to properly craft our dependencies.

Kind regards

        Andreas.


[1] https://github.com/bioconda/bioconda-recipes
[2] https://github.com/bioconda/bioconda-recipes/tree/master/recipes/qiime

-- 
http://fam-tille.de


Reply to: