Re: [AMBER-Developers] Dashboard CC builds all failing

From: Ross Walker <>
Date: Wed, 01 Oct 2014 10:13:07 -0700

Hi All,

The Cruise Control machine represents a 'standard' - read default, RedHat
6.5 build. This means that suddenly requiring non-standard (as in NOT
default) packages and libraries will break the build. Clearly this shows
that python-dev is not part of the default Redhat 6.5 install - this is
unfortunate - Ideally we should not be relying on such things and I would
suggest that we find a way to work around them. The more we rely on
external libraries and packages with AMBER the more difficult it is going
to be to keep things working on large numbers of different operating
systems. Just because it works on your bleeding edge Linux installation
does not mean it works on everything.

Remember in a LOT of cases people do NOT have root on the systems they are
trying to build AMBER on. Expecting them to install a bunch of external
packages is likely to be a real administrative pain in the butt for them.

Also, IBM Power systems (with AIX!!! OMG!!! :-( ) are about to start
appearing again in the systems people run AMBER on - so if you've added
the need for external packages to your favorite project please be prepared
to do the work to get this working on such a system. You have been warned!

So, Jason, before I install python-dev on the Cruise Control machine can
you please let me know exactly why it is needed and why it cannot possibly
be worked around to avoid the external dependency? If your arguments are
good, and enough people here are convinced then I'll add the package to
the cruise control machine. Otherwise we should probably set whatever
packages in AMBER require this to be 'optional' builds that are not
compiled by default.

All the best

On 10/1/14, 9:39 AM, "Gerald Monard" <>

>On my side I check "regularly" the Cruise Control (like every month)
>because there is some specific sebomd builds. From my experience, builds
>have failed quite repetively these last months because of changes in
>configure, librairies, and such. People committed change, builds failed,
>then corrections were committed, etc. This means that sometimes, when I
>see a build that fails, just waiting a few days is enough to make the
>build works again :-). I know, this is not a good behavior, and I should
>not wait for the others to repair problems (that I did not introduce).
>But I would guess that this is pretty a general behavior.
>Anyway, I still think CC is a great tool.
>On 10/01/2014 03:34 PM, Jason Swails wrote:
>> Hi everyone,
>> Are we still trying to enable/use continuous integration through Cruise
>> Control? Every build is now failing because the host machine does not
>> have the python-dev package installed. Dan added this protection to
>> configure back on Sep. 19, so CC hasn't made it past the configure stage
>> on any development branch since then.
>> CI is an awesome concept (and works great in projects where it is
>> adopted), but it doesn't seem to have caught on much with Amber...
>> All the best,
>> Jason
> Prof. Gerald MONARD
> SRSMC, Université de Lorraine, CNRS
> Boulevard des Aiguillettes B.P. 70239
> F-54506 Vandoeuvre-les-Nancy, FRANCE
> e-mail :
> tel. : +33 (0)383.684.381
> fax : +33 (0)383.684.371
> web :
>AMBER-Developers mailing list

AMBER-Developers mailing list
Received on Wed Oct 01 2014 - 10:30:02 PDT
Custom Search