Hi Sasha, On Wed, 2008-07-02 at 16:11 +0300, Sasha Khapyorsky wrote: > Hi Hal, > > On 14:28 Mon 30 Jun , Hal Rosenstock wrote: > > > > > > As we learned recently having structured API leads to such problems just > > > well, in this case one can expect "stable API" even if it was never > > > declared. > > > > I hardly think that situation is comparable if you are referring to the > > library APIs exposed inside of OpenSM for the vendors who wanted these > > without OpenSM itself (for diags and ibutils). > > Why not?
Because exposing all the data structures, etc. is way more granular (all the socalled opaque objects are no longer opaque) than a library API. It's a totally different magnitude IMO. > We got complains about changes in libosmcomp just few weeks > ago. But at least that's way more manageable the way it is and I would have cast the libosmcomp issue in a slightly different light. > I clearly don't want to make such limitations for OpenSM itself. I don't see how those limitations can be made or OpenSM could not move forward. > > > I think I agree with Ira - OpenSM version exact match enforcement will > > > make it clearer that plugin writer should be ready to rebuild and > > > possibly update its code. > > > > should being the operative word. > > > > > > Might this be better handled as packaging with separate packages based > > > > on licenses ? > > > > > > What do you mean? > > > > This is related to what I wrote below about the OpenFabrics licensing > > requirements. The idea is if GPL licensing were to be allowed (perhaps > > only in some limited context), then there could be two different > > packages: dual and GPL. In that way plugins would be more assured of > > being compatible with each other and OpenSM. > > I would prefer to separate packages by its functionality and not due to > licensing issues. Right, licensed based packages do put all related functionality in a similar bucket (management) but is that the basis to make such a decision ? > Anyway GPL is not permitted in OFA and all this > discussion is yet hypothetical. Yes, but one needs to look at hypotheticals to set direction. -- Hal > Sasha > _______________________________________________ > general mailing list > [email protected] > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general > > To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
