Boost Testing :
From: Beman Dawes (bdawes_at_[hidden])
Date: 2007-09-17 21:47:05
Victor A. Wagner Jr. wrote:
> Beman Dawes wrote:
>> Boost.System tests were added to boost-trunk/status on September 12th,
>> yet http://beta.boost.org/development/tests/trunk/developer/system.html
>> is still not reporting any results for those tests on the Sandia, OSL5,
>> and Rudbek Associates tests. From the dates for those tests, it would
>> appear those systems are not running on the current state of
>> boost-trunk/status, and that brings into question whether they are
>> updating their working copies at all.
>> What can be done to fix the immediate problem?
>> Why did the problem occur?
>> What can be done to ensure the problem does not reoccur?
>> What can be done to automatically detect and report the problem if it
>> does reoccur?
>> We need to start asking those questions every time there is any kind of
>> failure with the regression testing mechanism. Timely and accurate
>> regression testing is at the top of the list for improving the Boost
>> release process.
> this script runs continuously on my system
> C:\Projects>type svnboost.bat
> python regression.py --v2 --incremental --bjam-options=j2
> --user=anonymous --toolsets=msvc-7.1,msvc-8.0
> --runner=RudbekAssociates-V2 --monitored
> cd boost
> tools\jam\src\bin.ntx86\bjam --build-dir=c:\projects\results
> toolset=msvc-8.0 install
> cd ..
> sleep 1800
> if there's something than needs changed, someone has to tell us
Here is the message Rene posted Sunday:
> Last month, I mentioned that testers needed to *manually* update the
> regression.py script. Currently not having an up-to-date script is
> causing stale sources to be tested instead of the current state.
> ** Please update your regression.py script as soon as possible. **
> You can get the latest version of the script directly from SVN by
> downloading it
> If you can, I suggest inserting a step in your testing to update the
> script from that location. I personally use "wget" right before running
> regression.py so that I always have the most up to date version.
Presumably that's why the Rudbek tests still aren't reporting anything