Re: [Boost-testing] Significance of Timestamps

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2005-01-07 12:37:20

Stephen W. Carson writes:
> When I go to the regression report I see "Tue, 04 Jan 2005 16:43:53" 
> under my darwin run.
> I just kicked off another test (today, Jan 7) and I see "Getting sources
> (Tue, 04 Jan 2005 16:43:53)".

That's definitely wrong. 

>
> So I'm not sure what to make of these timestamps.  If, in both places, 
> it means the last time the sources were changed 

More precisely, the timestamp marks the beginning of the sources/tarball
retrieval.

> then why do I see 
> timestamps after Jan 04 on most of the other test runs?

Looking at the script, I think that our timestamp refreshing code
relies on a non-portable/unspecified behavior (we assume that doing
something as simple as "open( timestamp_path, 'w' ).close()" changes
file's modification time). Fixed in the CVS now; please grab the
update version from
www.meta-comm.com/engineering/boost/regression.tar.gz.

Thanks for reporting this,
-- 
Aleksey Gurtovoy
MetaCommunications Engineering