>But it's not good when you want to interoperate. You'll end-up having >code in your applications trying to do some black magic to decide what >format should it send the rkey to the remote side. IMO that's bad and >that should be fixed. Applications shouldn't have to be aware of these >details.
I agree. But things like DAPL exist for applications that don't want to be aware of OS specific details. I'm also proposing that the libibverbs, libibumad, and libibmad *interfaces* be ported over the Windows verb interfaces, which would make it that much easier to port apps between the different OFA products. >I believe each IB stack should give rkey in net order, and accept them >in net order. Changing the behavior of existing interfaces isn't really an option at this point. - Sean _______________________________________________ ofw mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
