Open Stack

On 10/12/13 13:34, Clint Byrum wrote:
> Excerpts from Steven Hardy's message of 2013-12-10 03:00:26 -0800:
>> On Tue, Dec 10, 2013 at 11:45:11AM +0200, Pavlo Shchelokovskyy wrote:
>>> - wouldn't it be better to keep all these changes in one bug and fix all
>>> misuses per file basis (with one file per patch-set for example)? It seems
>>> to me it would be easier to review in this way.
>>
>> One file per patch is not a good idea IMO, I think the review burden is
>> minimised if you make sure that each commit just contains the exact same
>> change to many files, then it's quick to click through them all and confirm
>> all looks OK.
>
> +1

FWIW I think I would have picked the other way. TBH any patch that 
contains dozens of lines of changes to 80+ files might as well not be 
reviewed at all, because the chance of any errors being spotted 
approaches zero. Hopefully everything will be done in an automated 
manner to minimise the possibility of that, and these are the unit tests 
so you'd expect a higher than usual chance of the gate failing if there 
is a problem. A change that touches just about every unit test file is 
also practically guaranteed to conflict with a high proportion of other 
patches going in, so it will be interesting to see if this makes it or 
gets stuck in an infinite rebase loop. By all means give it a go though :)

cheers,
Zane.