> > And the > > customers can pick the one(s) they like. > > I see your point regarding QA effort. Is it really twice? > Probably more - I'm reasonably sure most scripts written so far > assume stuff is installed in prefix/bin, so testing harness etc > would need to be changed. > And how to make sure the *correct* set of binaries was actually > QA'd? As far as I understood each component owner is responsible for QA of her/his component and supported platforms. For PPC we cover that. > > If we have 32bit execs only for development resp. testing, why don't we > > also give them to customers in order to do basic test or diagnosis of > > their setup? > Because of the confusion this would create. > For shared libraries, the 32/64 bit issues seem to be automatically > figured out > by ld.so, but there's no such solution for binaries. This is true as we have discussed at ofed-1.1. See also http://openib.org/pipermail/openfabrics-ewg/2006-October/001831.html I agree with you in that if there is a standard for binaries dir struct let's go for it. If there is no such one, let's agree on one approach: either bin32 resp bin or appl resp appl64 or... To me it's worse if customers have to fix or write build scripts by themselves in order to build 32bit binaries.
Nam _______________________________________________ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
