On Wed, 2008-05-14 at 00:15 +0000, Sasha Khapyorsky wrote: > On 13:56 Tue 13 May , Hal Rosenstock wrote: > > > > > > That is correct, but we know that ibsim parser doesn't parse > > > NodeDescription in those (port related) lines, so in such "worst" cases > > > when 's=' and/or 'w=' strings are used in NodeDescription this could be > > > just filtered out from ibnetdiscovery file. > > > > That's why I termed this approach a workaround and it does limit the > > NodeDescription in ways not limited by the IBA spec. > > No, it does not limit NodeDescription at all - it is *only* file format > limitation (remove NodeDescription from port related lines in the file > and we are done).
That's my point; NodeDescription is part of the ibnd file format and hence this limits what can be there so in that sense this limits it's contents and removing such occurrences from the file is a workaround IMO. While a distinction can be made between the actual NodeDescription and what is in the file, they're supposed to be one and the same and not require some extra post processing. > > Is this worth mentioning in the README or some other doc for ibsim ? > > Looks like overkill for me. OK. -- 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
