On 04/19/2016 12:30 PM, John Ferlan wrote:
On 04/19/2016 11:10 AM, Cole Robinson wrote:
> On 04/19/2016 08:01 AM, John Ferlan wrote:
>>
>>
>> On 04/18/2016 05:48 PM, Cole Robinson wrote:
>>> On 03/02/2016 01:55 PM, John Ferlan wrote:
>>>> Move to secret_conf.c and rename to virSecretLoadAllConfigs. Also
includes
>>>> moving/renaming the supporting virSecretLoad, virSecretLoadValue, and
>>>> virSecretLoadValidateUUID.
>>>>
>>>> Signed-off-by: John Ferlan <jferlan(a)redhat.com>
>>>> ---
>>>> src/conf/secret_conf.c | 175
+++++++++++++++++++++++++++++++++++++++++++++
>>>> src/conf/secret_conf.h | 3 +
>>>> src/libvirt_private.syms | 1 +
>>>> src/secret/secret_driver.c | 174
+-------------------------------------------
>>>> 4 files changed, 181 insertions(+), 172 deletions(-)
>>>
>>> ACK, mirrors network_conf.c layout.
>>>
>>> (though honestly I'd rather we have separate files for XML handling and
object
>>> handling... the existing conf.c files are too large anyways. I put an entry
on
>>> the LibvirtFirstBugs wiki page about that type of code reorg though it's
>>> probably over an initial contributors head)
>>>
>>
>> I could go the route of a "virsecretobj.c" (to partially mimic the
>> virdomainobjlist.c) instead of cramming everything into secret_conf -
>> there's about 1000 new lines in secret_conf just for this series (with
>> the reduction of lines in secret_driver).
>>
>> It would mean a complete v2 of this series - although considering most
>> things have been ACK'd the second review should be easier. I think
>> there's only a couple of issues/questions to resolve from reviews.
>>
>
> Oh I wasn't trying to imply making it a requirement of this series. Definitely
> an additive thing, and I'm not even trying to pin you with the work, just
> floating the general idea
>
Understood - I'd rather do it now than take a hit at some unknown future
point in time to essentially do the same thing. In the long run it's the
"right thing" to do.
Yeah but I hate to see self contained changes get held up just to accommodate
an additional cleanup. Reviewer bandwidth is the bottleneck here
- Cole