Regression test results called into question

From: Victor A. Wagner Jr. (vawjr_at_[hidden])
Date: 2004-10-29 13:58:37

At Friday 2004-10-29 11:05, you wrote:
>At Friday 2004-10-29 08:20, you wrote:
>>On Fri, 29 Oct 2004 00:20:29 -0700, Victor A. Wagner Jr. wrote
>> > At Thursday 2004-10-28 23:42, Victor A. Wagner Jr. wrote:
>> > >I got tired of getting the exceptions so I fixed time_formatters.hpp 
>> which
>> > >had a logic error in it.
>>
>>I don't see a change in CVS -- please send it to me and I will verify, test,
>>and check in. I'm suprised you found something, b/c we've gone looking for
>>the cause of this before and come up empty...
>>
>> > >It's possible that the test should now be run on all platforms again.
>> > >Since I don't know how to alter which tests actually run on which
>> > >platforms I cannot fix that.
>> >
>> > so I run the regression script I have.... it doesn't rebuild the
>> > vc8.0 teststreams.exe what's up?
>> >
>> > _surely_ I don't have to figure out what tests are affected by an
>> > include file change
>>
>>I don't think that the bjam dependency finding is 100% perfect. Just delete
>>the executable -- that will force the rebuild.
>
>That would seem to be a very serious problem since the regression script 
>doesn't automatically delete all the .exe also it doesn't compile to 
>the same .exe I get when I go directly to the test directory and compile 
>with VS.net2005 beta (Whidbey)

deleting the .exe didn't help
apparently one has to delete the .obj files

Right now it is my belief that all the regression test results we see are 
extremely questionable.
We have NO idea what is actually being tested.

Absent some change in the testing, I certainly wouldn't sign off on a release.

>>Jeff
>>
>
>Victor A. Wagner Jr. http://rudbek.com
>The five most dangerous words in the English language:
> "There oughta be a law"