Windows build time: 50% longer....

[Stephan Bergmann]

On 06/17/2015 09:31 AM, Norbert Thiebaud wrote:
> On Wed, Jun 17, 2015 at 2:10 AM, Stephan Bergmann <sbergman at redhat.com> wrote:
>> On 06/17/2015 01:58 AM, Norbert Thiebaud wrote:
>>>
>>> most likely candidate is prolly:
>>>
>>> 9b06d36571bbfa2d722f36610375e7cd24be81a7 reenable
>>> JunitTest_dbaccess_complex part RowSet
>>
>>
>> ...which should be easy to verify with "time make
>> JunitTest_dbaccess_complex" vs. "git revert
>> 9b06d36571bbfa2d722f36610375e7cd24be81a7 && time make
>> JunitTest_dbaccess_complex", no?
>
> Not when you have a build queue of 25 jobs, and are scrambling to
> bring some ressource online to try to drain that queue before people
> wake up and pile up again....
>
> iow 100% of the Windows I have access are working hard to keep up... I
> do not have spare to do this 'easy' thing....
>
> I've narrowed it down to about 20 commits using build history data..
> that, I could do without stopping builders....

for my random Windows 64-bit --enable-dbgutil etc. build, time values as 
per above are

   real 2m26.160s
   user 0m0.000s
   sys  0m0.000s

vs.

   real 0m46.872s
   user 0m0.000s
   sys  0m0.000s