Boost logo

Boost Testing :

From: Boris Gubenko (Boris.Gubenko_at_[hidden])
Date: 2005-12-10 07:42:31

Markus Schoepflin wrote:
> I bet you don't use CVS when running the boost regression tests.

No, I don't.

> I'm guessing that there is no such tarball
> (boost-Version_1_33_1.tar.bz2) on the source code server, [...]

You are right. When I tried yesterday to unzip the bz2 file manually,
I've discovered, that there is no such file in the current directory.
"not a bzip2 file" error in this situation is a bit misleading :-)

> What you could do is download the boost release, and run the
> regression
> tests with that. This is supported somehow by the python script,
> but I'm
> unsure of the exact details.

What I was doing on the TestDrive machine until I built python with
the bz2 module was:

I would do get-source on another machine:

   python get-source --runner=...

and, then, would do setup followed by the test on the target machine:

   python setup --runner=... --toolsets=... ...
   python test --runner=... --toolsets=... ...

The machines shared the disk, of course.

This sequence seemed to work, but if somebody can confirm, that this
is the right way of doing things it will be appreciated. The test
command does not upload results to the metacomm site, but for what
I'm doing now it is not needed.

What I'm trying to do is to run boost regression tests as part of the
compiler regression testing. This is why I need a "stable" version of
boost as opposite to CVS-HEAD. I hoped, that I can do it with just
one python command, pretty much like with CVS-HEAD, but if there no
tarball in CVS for 1.31.1, I will download and unpack it manually and
will do the rest with individual python commands.

> You could also try to use anonymous CVS when running the regression
> tests, then you don't need the tarball at all.

You mean --user="anonymous" --tag=Version_1_33_1 ? Do I need to install
CVS client for this?

Thanks for the help!


Boost-testing list run by mbergal at