Re: [Boost-testing] regression testing for vc-7_1

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2005-07-18 16:20:14

Stephen W. Carson writes:
> Aleksey Gurtovoy wrote:
>
>>Most likely you've happen to come across the issue of tarball
>>publishing being a non-atomic operation
>>(http://thread.gmane.org/gmane.comp.lib.boost.testing/1123). We'll
>>move this up in our TODO list, but meanwhile, please just give it
>>another try.
>>  
>>
>     Please do. I can't tell you how many test runs have failed because 
> of this. I usually kick tests off at night and either forget to or don't 
> have time to watch and see if the tarball problem is hit. Then I check 
> in the morning and the whole night has been wasted! Very frustrating.

Well, it's good to know. We've had no idea about the extent of the
problem. The tarball creating procedure is now fixed to be as close to
an atomic operation as possible, and 'regression.py' is modified to
retry if the tarball download/logs upload was unsuccessful.

-- 
Aleksey Gurtovoy
MetaCommunications Engineering