On 04 Nov 2013, at 18:35, Fotis Georgatos wrote: > > Hi *, > > On Nov 4, 2013, at 5:13 PM, Kenneth Hoste wrote: >> Suggestions for other (broad) topics are welcome, I'll add them to the conf >> call description and wiki page >> (https://github.com/hpcugent/easybuild/wiki/Conference-calls). > > One topic I care about ASAP, is the goolf vs goolfc debate described over > here: > https://github.com/hpcugent/easybuild-easyconfigs/issues/488 > I'd like to hear developers' feedback as regards current standpoint, > in order to take course with the relevant open PRs.
There's nothing much to discuss here: goolfc is a superset of goolf, so they're compatible (mixing modules there is not an issue), and -malign-double is enabled by default on x86-64 (see http://gcc.gnu.org/onlinedocs/gcc/i386-and-x86_002d64-Options.html)... See also comments in the mentioned PR> > Also, I have a pile of topics, which I rather maintain as personal itches, > yet others may care about; the first 3 are fairly generic subjects: I'll let others decide which points should/will be discussed in the conf call. There won't be time for everything though... Also, I'd like to see issues opened for these before discussing them in a conf call, We need a handle to track relevant remarks/opinions, mailing stuff back and forth will only lead to stuff that's out-of-context and forgotten... Fotis: please open issues for any of the below that don't have an issue yet, and then they're eligible for the conf call. Also, please specify which one of these you would like to add to the agenda (before 2pm, let's say). > * toolchain-less builds; important for debuggers, performance tools and all > non-ABI exposing software > * module statistics! # how to record module usage; what do you do? > * Define a few more example ModuleNamingSchemes, such as: "lowercase", > "SARA", "TAMU", "FZK" etc > * #260: Avoid long paths in `eb --search` and `eb --dry-run`, trim what is > visible in $CFGS > * parallel builder to understand "blocker" items (avoid delivering icc + impi > + imkl in parallel) > * Inform upstream developers of patch effort, add REF. with tracking URL > within patch files > * per site customization effort (see EB-F#712, #544), .ebp files or, EB_EC_* > vars > * automate eb --try-toolchain=X -r > * EasyDoc/VP; https://hpc.uni.lu/users/software/ > Why? check links > http://www.fz-juelich.de/ias/jsc/EN/Expertise/Support/Software/Libraries/PETSc.html > * EasyRegTest/XB. oh-mpi-god. Why? > http://my.cdash.org/index.php?project=HPC-OpenMPI > * FFTW2 vs FFTW3 vs imkl; API divergence in ictce & g???? toolchains > * diverse build regtest environment, ensure multiple OSes/archs/distros tested > > I'd prefer others choose if/which ones they think would matter for all of us; > none of them is critical yet, from my point of view. Feel free to postpone > the topics for a next beer round, just invite me for the beers :)
