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

Re: Refs to Conda - giving it a try after meeting Björn at a conference

On Sun, Sep 22, 2019 at 10:44:28AM +0200, Steffen Möller wrote:
> > We have some things in progress with Björn and other Elixir project
> > members to link tools with the github/bio.tools stuff. Things are going
> > slowly for some time. I am involved as part of biocontainers with Björn.
> I recall to have asked before about adding refs to conda directly. The idea
> back then (half a year ago?) was to not compete with what bio.tools should
> be providing. I mean - there is only one place that can attract the
> eyeballs.
> And it saves us some work, too :o)

I admit I do not understand this remark (and its effect to adding the
conda:bioconda references).
> But it is not so bad. I am at "j" now

Thanks a lot for your busy work on this.

>  - the task pages are seriously
> behind, which is another issue.

I did not even announced it working at all. ;-)
Well, I managed to get it working.  I need to repeat:  Once you changed
some XML you **really** should run xmllint.  I've fixed the syntax error
of the template and for sure I also needed to edit other files (please
remember: It is NOT ONLY the template that needs to be changed.  The
according properties need to be set first inside web sentinel Python
code as well as the database query.  So may be you ask me next time -
its also rarely enough that it needs to be done to write an extensive
documentation about it.

However, the reason that its lagging behind is that you managed to
break the UDD importer again.  Here is the fix:


I told you that the UDD importer is quite sensible and I can only catch
issues that happened before.  Since never before any Registry came
without a Name field that was not catched.  It is now but I urgently
ask you to *really* *check* what you are commiting.  Please read
twice when commiting issues like I fixed here:


or if you are not sure use yamllint.  It does not break the UDD
importer completely any more but finding those issues later always
costs more time (for you or for your fellow team members).

> > won't be there, I have an other meeting same date (though in Paris too...)
> Ok. Andreas is attending.
> I just wanted to fix the CSS of the bioconda links but when I found
> where to change it, I also saw that Andreas was already at it. So, many
> thanks for that, and at some point we should talk about our
> infrastructure a bit more.

I hope it is fixed now.  I guess my browser cache is providing me with
the old css.  Can anybody confirm that the "bioconda" links are
unerlayed with some lightgreen color?

> The delays affect also the motivation to
> update a package.

As I said: Please keep on updating - but slower and with correct syntax.
> Sidenote - Bjoern explained how they have a bot for routine updates that
> creates a pull request for human inspection immediately upon spotting a
> new version, which also summarizes the result of the autotests. That is
> lovely. We should come up with something like that - final decision with
> the human, automated what can be automated.

That "we should come up" is a magical "we".  "We should also care for
Python3" conversions and "we should care for autopkgtest" while "we
should package new stuff".  I think it would be better if somebody
would subscribe to a certain task like "*I* will work on xyz" since
that "we" is so anonymous that nothing will happen.

Kind regards



Reply to: