Hi, apologies to jump late in this thread, but I just noticed this part:
> along the lines of xz, gzip and such IMHO, this request is not as much a case of #1689 (the objective there is to prune the combinatorial space of deps of future builds, to ease collaboration) https://github.com/hpcugent/easybuild-easyconfigs/issues/1689 as much as it is this kind of thingy: http://hpcbios.readthedocs.org/en/latest/HPCBIOS_2012-90.html N.B. that policy DOES NOT specify if individual components come via packages or modules - that’s very intentional. Furthermore, the selection of MUST/MAY/SHOULD keywords make it very lightweight and compliance with it is quite easy. In effect, all that an HPC site has to do is to document precisely which “Subcategories” are of interest/provided. That document is surely not cast in stone and once we get a chance to run the policy targets against multiple distros (rhel5, rhel6, rhel7, debian6, debian7, ubuntu*) it will be refreshed. Ideally, there should be ready templates for many configuration management platforms (cfengine, puppet, ansible etc) so that the path of least resistance is very obvious.. Sysadmin should have choice to deliver either via packages or via easybuild processes. This makes lot of sense at an HPC site with multiple distros. Any feedback here? to be continued, F. On Jun 19, 2015, at 1:01 PM, Martin <[email protected]> wrote: > The issue seems to be exactly what I'm talking about. I'll take back anything > that has any implications for foss/2015b and start confusing people in > issue#1689 (once I read up on it) :) > > On Fri, Jun 19, 2015 at 1:10 PM Kenneth Hoste <[email protected]> wrote: > Hi Martin, > > > On 19/06/15 12:05, Martin wrote: >> Hello, >> >> I'm not sure wether it should be part of the toolchain but to me "base >> tools/libs" should somehow be standardized on. I'm thinking along the lines >> of xz, bzip and such (I'm more along the lines of: Provide an environment as >> complete as possible for the users). Most of the time these tools are not >> technically required but I found that a lot of users expect the tools to >> simply be available and in a working state. >> >> If that opens a a completely different can of worms please feel free to >> delay this topic until foss/2015b is done and another solution can be >> discussed. > > This issue may be relevant, depending on what you are looking for exactly... > > https://github.com/hpcugent/easybuild-easyconfigs/issues/1689 > > > regards, > > Kenneth > >> >> /Martin >> >> >> On Fri, Jun 19, 2015 at 10:26 AM Kenneth Hoste <[email protected]> >> wrote: >> Hi Christopher, >> >> On 19/06/15 01:20, Christopher Samuel wrote: >> > On 17/06/15 20:50, Kenneth Hoste wrote: >> > >> >> OpenMPI v1.8.5 (note: with 1.8.6 likely to be released very soon, >> >> this may get bumped before the definition is cast into stone) >> > 1.8.6 came out the night before last. :-) >> > >> Yes, I noticed too, I've already bumped OpenMPI to 1.8.6 in the >> gompi/2015b and foss/2015b PRs. >> I also bumped hwloc (a dep for OpenMPI) to 1.11.0. >> >> >> regards, >> >> Kenneth >> -- >> -- >> http://www.xing.com/profile/Martin_Marcher >> http://www.linkedin.com/in/martinmarcher >> Mobil: +43 / 660 / 62 45 103 >> UID: ATU68801424 > > -- > -- > http://www.xing.com/profile/Martin_Marcher > http://www.linkedin.com/in/martinmarcher > Mobil: +43 / 660 / 62 45 103 > UID: ATU68801424 cheers, Fotis -- echo "sysadmin know better bash than english" | sed s/min/mins/ \ | sed 's/better bash/bash better/' # signal detected in a CERN forum
