On 04/14/2011 11:42 AM, Michal Novotny wrote:
[snip]
I've fixed those OOMErrors and the whitespace already for v3 however I
won't send the v3 of the patch until I get reply to this since some
other approach may have to be chosen for the port = 0 things. comments
inline ...
> Hmm, why exactly are these port = 0 changes needed? Were they breaking
> some tests?
Exactly. It was breaking almost all of the xmconfigtests and
sexpr2xmltests. The reason was that it was never set in the code to
setup just one serial/parallel port and this is why it was breaking it.
That's the reason for those changes to port = 0.
Yeah, at first I didn't think they were necessary, but I see now that
they are. Reason being that we only do the port allocation when parsing
XML, not when formatting any random DomainDefPtr like I was assuming. So
it's better to have all the callers manually set a target port if
building a ChrDefPtr by hand. Sorry for the noise.
Thanks,
Cole
[snip]
> I think these test files would be better named serial-target-port-auto.
Good, I'll do that name change for v3.
Thanks,
Michal