Re: [boost] [Report] 1668 regressions on RC_1_34_0 (2007-01-27)

From: Joe Gottman (jgottman_at_[hidden])
Date: 2007-01-27 23:43:07

Douglas Gregor wrote:
> Boost Regression test failures
> Report time: 2007-01-27T13:27:45Z
> 
> This report lists all regression test failures on release platforms.
> 
> Detailed report:
>   http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/issues.html
> 
> The following platforms have a large number of failures:
>   borland-5.8.2
>   msvc-6.5
> 
> 1668 failures in 12 libraries (49 are from non-broken platforms)
>   algorithm/string (1)
>   date_time (0 of 14 failures are from non-broken platforms)
>   dynamic_bitset (0 of 1 failures are from non-broken platforms)
>   graph (1)
>   iostreams (1)
>   multi_index (0 of 2 failures are from non-broken platforms)
>   regex (36 of 38 failures are from non-broken platforms)
>   serialization (0 of 1582 failures are from non-broken platforms)
>   test (1 of 19 failures are from non-broken platforms)
>   tr1 (4)
>   type_traits (2)
>   wave (3)

    I can't help noticing that 1582 of 1668 regressions are in the 
serialization library.  That's almost 95% of the regressions.  And the 
next highest number of regressions is only 38.  What is it that makes 
the serialization library so fragile?

Joe Gottman