Thank you guys for your replies. I'm happy to see I'm not the only one that 
would benefit from this.

I totally agree with Kenneth on how this should be implemented. Am I wrong to 
assume that a positive side effect of this approach is to make the 
"hiddendependencies" option for easyconfig files redundant since EB will 
automatically look for hidden dependencies if no visible ones are found for the 
ones listed in "dependencies" and "builddependencies"?

DV

--
Davide Vanzo, PhD
Application Developer
Adjunct Assistant Professor of Chemical and Biomolecular Engineering
Advanced Computing Center for Research and Education (ACCRE)
Vanderbilt University - Hill Center 201
(615)-875-9137
www.accre.vanderbilt.edu

On Aug 4 2016, at 5:11 am, Kenneth Hoste <kenneth.ho...@ugent.be> wrote:

Hi Markus,

On 04/08/16 10:31, Markus Geimer wrote:
> All,
>
> It seems that with [2] in place, the dependency resolution already
> does "the right thing" as it parses the easyconfig files. But that
> obviously only works if the easyconfig contains the 'hidden'
> parameter. That said, maybe it is sufficient to ensure writing the
> 'hidden' info (wherever it comes from: 'hidden' flag, 'hide-deps',
> '--hidden' command-line option, ...) to the archived easyconfig?
> Or am I missing something?
That's probably only a partial solution, since you have no guarantee
that the archived easyconfig will be picked up the next time the same
installation is needed (the path to the archived easyconfigs is not in
the robot path, by default).

I *think* that making the robot consider also hidden modules is as
simple as updating the 'find_resolved_modules' function to optionally
check also whether the module is available as hidden (which involves
tweaking the dep_mod_name to inject a '.', or something like that).

You may need to make a similar change to make dry-run spit out the
correct info though, which probably means some refactoring to ensure
that both the robot and dry-run use the same function to check whether a
module is available or not...

regards,

Kenneth

>
> Markus
>
>
> [2] https://github.com/hpcugent/easybuild-framework/pull/1837
>
> On 08/04/16 08:58, Markus Geimer wrote:
>> Davide,
>>
>>> The most logical way of doing so seems to be by creating hidden
>>> modules for the dependencies we don't want the users to see. However,
>>> when installing other easyconfig packages from the default easyconfig
>>> files they cannot see the hidden modules and try to install them again.
>>> Is there a way to tell EB to automatically look for hidden packages
>>> without modifying the easyconfig files?
>> At this point not, unfortunately. A corresponding issue has
>> been created already quite some time ago [1], but since then
>> hasn't received much attention. Since I'm very interested in
>> this feature as well and would like to see it implemented
>> rather sooner than later, I recently started looking into it.
>> But I'm still far from a clean and working solution -- though
>> Kenneth claims that it shouldn't be too hard... But that's
>> what he is saying all the time ;-)
>>
>> Markus
>>
>>
>> [1] https://github.com/hpcugent/easybuild-framework/issues/1079
>
> --
> Dr. Markus Geimer
> Juelich Supercomputing Centre
> Institute for Advanced Simulation
> Forschungszentrum Juelich GmbH
> 52425 Juelich, Germany
>
> Phone: +49-2461-61-1773
> Fax: +49-2461-61-6656
> E-mail: m.gei...@fz-juelich.de
> WWW: http://www.fz-juelich.de/jsc/
>
>
>
> ------------------------------------------------------------------------------------------------
> ------------------------------------------------------------------------------------------------
> Forschungszentrum Juelich GmbH
> 52425 Juelich
> Sitz der Gesellschaft: Juelich
> Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
> Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
> Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
> Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
> Prof. Dr. Sebastian M. Schmidt
> ------------------------------------------------------------------------------------------------
> ------------------------------------------------------------------------------------------------
>

Reply via email to