[ovs-dev] [PATCH v3 2/8] Test stack accross patch port.

[Jarno Rajahalme]

> On Mar 16, 2015, at 3:07 PM, Ben Pfaff <blp at nicira.com> wrote:
> 
> On Fri, Mar 13, 2015 at 04:51:55PM -0700, Jarno Rajahalme wrote:
>> This added test should fail. Next patch fixes OVS to pass this test.
> 
> Thanks for adding tests!  More tests now mean easier work later.
> 
> It's not nice to have a commit that breaks tests, even if it's fixed
> by the following commit; it can cause trouble in bisecting, for
> example.  So, I would usually add a commit like this just *after* the
> commit that fixes the problem, or combine it with the commit that does
> the fix.
> 

OK, I’ll squash this with the following patch.

>> +AT_CHECK([ovs-appctl time/warp 500], [0],
>> +[warped
>> +])
>> +sleep 1  # wait for log writer
> 
> I assume that usually the log write would be done much more quickly
> than 1 second later?  Is it easy to replace this "sleep" by an
> OVS_WAIT_WHILE or OVS_WAIT_UNTIL?  It is nice to skip sleeps, if one
> can.
> 

I’ll use this instead:

OVS_WAIT_UNTIL([test `grep flow_add ovs-vswitchd.log | wc -l` -ge 1])

>> +# The WARN was intended, remove it
>> +rm ovs-vswitchd.log
>> +touch ovs-vswitchd.log
> 
> It would be better, if it is possible, to instead skip this warning
> via the argument to OVS_VSWITCHD_STOP.
> 

Thanks for the reminder, I’ll use this instead:

OVS_VSWITCHD_STOP(["/Failed to pop from an empty stack/d"])

>> +OVS_VSWITCHD_STOP
>> +AT_CLEANUP
>> +
>> AT_SETUP([ofproto-dpif - port duration])
>> OVS_VSWITCHD_START([set Bridge br0 protocols=OpenFlow13])
>> ADD_OF_PORTS([br0], 1, 2)
> 
> Thanks,
> 
> Ben.