Re: [boost] [Report] 1472 regressions on RC_1_34_0

From: Roland Schwarz (roland.schwarz_at_[hidden])
Date: 2007-01-15 12:55:55

Jeff Garland wrote:
> Seems like most of the new ones are bogus -- here's one from date-time that 
> looks like it ran just fine, but something is wrong with the XML that's 
> 'causing the failure' -- as you can see, the test actually passed.

I have to say that I am starting to doubt that the regression result 
pages are of much value at all. I am trying to go for some of the bugs, 
but more often than not am not able to reproduce what I am seeing on the 
reports page :-(

Perhaps I am doing something ill advised, but the regression.py local 
runs are strange. I often find the regression hang in a dialog box 
waiting for some confirmation. Is there a "hidden" switch to control 
this? How do the regression managers manage this? Or is this the reason 
that e.g. metacomm is stalled for over 10 days?

Another example: running the tests for msvc-6.5 for release (yes I know 
its old) by hand pass, but running them from regression.py gets the 
filesystem operations_test hang in an endless loop:
0040A7D0 jmp 0040A7D0
(Verified by attaching with a debugger)

I feel like the current situation does not allow to get a clear picture 
about the real "health" of the boost library.

How can I help improve on this?

Roland