Re: [Boost-testing] Tarball testing

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2004-11-02 17:29:07

Bronek Kozicki writes:
> Aleksey Gurtovoy wrote:
> > 1) Download the updated "regression.py" from 
> > 2) Download the archive of your choice:
> > 3) Place the script and the archive in the same directory (e.g.
> >    the same one you are currently using for tests on the branch)
> > 4) Modify the script command line to include '--local' option:
> > 5) That's it.
> > 
> 
> Three questions:
> 
> 1. it seems like this procedure requires running regression tests 
> manually (or at least with some tool like wget), right? 

Well, kind of; it's just one-time thing -- you download the tarballs
manually, run the tests once, and you are done until the next version
of tarballs is available (if any).

> What about tests I'm still running in background , ie.
> python.exe regression.py "--runner=Bronek Kozicki" --tag=RC_1_32_0 
> ---local=boost_1_32_0.zip --toolsets=vc-7_1,mingw-3_3_1,mingw-3_4_2 
>  >boost_regression.log 2>&1
> 
> are these still useful?

Generally, yes, but only if they don't take away resources from tarball
testing (they do for us, since our current cycle is ~11 hours).

> 
> 
> 2. what about --tag=RC_1_32_0 option? Should I keep it when running this 
> "manual regression test procedure" or remove it?

Doesn't matter, --local overrides it.

> 
> 
> 3. how often these manual regression tests should be run?
> 

Just once for every new tarballs set.

Let us know if you have more questions about this.

--
Aleksey Gurtovoy
MetaCommunications Engineering