[libvirt] patch option needs clarification

[Gene Czarcinski]

On 10/22/2012 04:32 PM, Laine Stump wrote:
> On 10/22/2012 04:25 PM, Laine Stump wrote:
>> On 10/22/2012 03:59 PM, Gene Czarcinski wrote:
>>
>>> Oops.  But then I am not sure how to do it.  The new code changes
>>> "everything."
>>>
>>> If you want, I can go back and remove it ... make yet-another-patch on
>>> top of the one one to way to the list which removes interface= from
>>> the conf-file and then submit a third patch which puts it back in.
>> I think we're crossing wires. This is what patches I think should be sent:
>>
>> 1) a patch to add --interface to the commandline
>>
>> 2) a patch to switch from using the "long commandline" to using a conf
>> file (which will still put the equivalent of --interface=xxx into the
>> conf file).
>>
>> Isn't that what you already have?
> Ah, I just saw that you've already sent the patch, and it *wasn't* on
> top of the patch that adds --interface. An alternate path would be to
> have the "switch to conf file" patch first (but *not* adding the
> --interface option), then remaking that patch to only add to the conf
> file (ie to be applied *after* this patch). Either way, we need to have
> them in two separate patches.
>
>
That first patch was crap and I wish I could have retracted the message 
after I sent it.

"v2" of the patch is "on the way" if not already posted.  This patch was 
created by:

1. checkout master; pull, checkout -b gc-cf-4 master
2. Using "patch -p1", apply my patch (no interface=) for 
bridge_driver.[ch], and networkxml2argvtest.c
3. Edit bridge_driver.c to fixup the things that did no go on clean.
4. create tarball from another tree for the *argv testfiles and untar 
onto gc-cf-4
5. commit and the format patch
6. create another branch from master and apply the created patch; fixup 
a couple end-of-line whitespace problem; reapply ... this time clean.
7 send-email

Well, one good thing is that I am starting to get the hang of git ;)

This version of the patch did not screw things up in bridge_driver.c 
like the last one did.

Yes, this version does not  include the "interface=" code which involves 
a couple of lines in bridge_driver.c plus an update to each of the argv 
test files.  A patch adding "interface=" to the previous patch will be 
submitted shortly.

I would appreciate an explanation why there is reluctance to adding 
"interface=".

Yes, there were problems a while ago if it was used, but there is now a 
definite problem if it is not specified and dnsmasq =>2.61.

Without "interface=", the bind-interfaces does not work and v4 and/or v6 
packets can be mis-routed by the kernel when there are multiple 
instances of dnsmasq running. Dnsmasq listens to 0.0.0.0:67/68 for v4 
and :::547 for v6.  Without good packet routing results are unpredictable.

Gene