Re: [Boost-testing] Re: range\config.hpp and regression tests broken

From: Martin Wille (mw8329_at_[hidden])
Date: 2005-04-17 10:36:29

John Maddock wrote:
>> I have deleted a couple of test results which looked outdated for 
>> various reasons. So, the symptoms should go away for now. However, the 
>> cause stays unknown (and I currently don't have enough time to 
>> investigate this thoroughly).
> 
> 
> The problem is still there, actually two problems:
> 
> unicode_iterator_test is failing at runtime with gcc-2.95.3, but passes 
> when run locally on my machine (I'm fairly sure that the error shown may 
> have been a real issue at some point, but shouldn't be now IMO).

The test binary for unicode_iterator_test/gcc-2.95.3 and the associated 
XML file on my Machine date back to 2004-02-15. Why doesn't it get 
rebuilt when there was a code change to fix the errors? Could this be a 
result of a failure in building some depend code which doesn't cause the 
  XML file to get deleted (that's the only idea I have)?

> unicode_iterator_test is shown not building with several compilers, due 
> to a Boost.Test change that was fixed a couple of months back (this was 
> the original cause of my last message).  These failures do not show up 
> on the old style test results, only the new style ones.  I suspect there 
> must be some old/stale xml results sitting on Martins hard drive that 
> are confusing the postprocessing (at Martins end that is).  Martin can 
> you try deleting the whole regex build-tree and see if that clears 
> things?

I'll delete the build- and the test-tree after the current run.

   If you've already done this recently, then obviously the cause
> must be elsewhere: are the xml results placed anywhere else by the 
> python driver script?

I have only deleted the results for some of the tests. I didn't see that 
this one is outdated, too.

I don't use the python script for running the tests (I'll use it when I 
set up tests on another system). For collecting and uploading the 
new-style results I use one of the python scripts. However, since the 
old binary and the old XML file both exist in my filesystem, the error 
likely is somewhere in the build-system or its testing subsystem.

Regards,
m