From: John Maddock (John_Maddock_at_[hidden])
Date: 2000-10-15 06:25:33
>Now to the bad news: some regression tests running failed.
>type_traits_test: 328 tests completed (43 failures)
I'm surprised that there weren't more failures!
Seriously though, I know of no compiler that gets everything correct in
there (and they all fail on different things), however most of the failure
are wierd corner cases; with the possible exception of VC6 builds, sorry
but without partial specialisation workarounds only go so far...
>I suggest to have some standardized output from test programs, so not only
>checking but also failure reporting could be automated. Using three
>compilers I really don't like to read all the test output by myself every
>time I checkout the cvs.
It would be good if the tests could be run as well as compiled, and be able
to return a persentage success where appropriate - ideas?
>btw, is cvs content usable for production or it is a bleeding edge?
For libraries that have been accepted into boost then in theory yes, there
may be some things in the cvs tree though though that have not reached
release yet (boost/pending and develop directories).
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk