Re: More cleverer QA config needed? (was Re: [pcp] NSS/NSPR Testing Status)

[Dave Brolley <brolley@xxxxxxxxxx>]

On 11/28/2012 09:34 PM, Nathan Scott wrote:
> Also, are you seeing this one fail Dave ... (nssmerge branch)? [4%] 
> 024 4s ... - output mismatch (see 024.out.bad) 2a3,4
> > __pmSetDataIPC: fd=<n> data=0xf83890
> > IPC table fd(PDU version):
[ ... ]
> I'd expect that to be on a new-fail list?  Needs treatment like
> 041 got recently.
I get

024 [not run] ./getpmcdhosts: unable to get 1 host(s) with options: "-L 
-n 1 -a sample"
for that one. However based on how similar tests were fixed, I have 
pushed a fix to brolley/nssmerge in the pcpfans repository.
> This actually raises another interesting twist on the QA config
> issue - this passes for me with some PCP_VER_3611 variants (ie.
> built --without-secure-sockets) but fails otherwise.
Yeah. __pmSetDataIPC is currently only used by the secure sockets 
implementation.
>    Suggests
> we need to consider doing something more sophisticated for the
> QA tests with NSS/NSPR dependencies - pkg-config(1) for PCP?
Can you elaborate? I'm not sure what you're getting at.

Dave