David Abrahams wrote:

Are you really testing CW 7.x?

Hmm - I thought I was. Actually I'm not doing the testing. I'm just relying on the table of test results which refers to compiler "CW" So I don't really know what version it is. CWPro9 fails export and wide char tests on my machine when built for debug and CWPro8 fails tests on Rene's machine and he builds for release. So I just assumed the CW referred to some previous version. Since it was passing I never had to find out what it really was.

If you really count on the result of is_convertible being true when it should be, then no, you don't have enough tests.

FWIW, I dropped CW 7.x support in Boost.Python and the iterators library when CW 9 came out and have not received a single complaint. I think VC6 is arguably the only 2-version-old compiler that might warrant continued support.

I never set out to make CW pass. It just did. I have spent time with CWPro as Rene expressed an interest and has been willing to help. The same goes for Borland where Pavel Vozenilek invested most of the effort. I wouldn't have done either of these were it not for this. I should say they did take a lot more effort than I originally anticipated - but then that applies to just about everything I do. Generally, I would agree that it makes no sense to spend effort to accommodate a compiler for which an upgrade is available. I wonder if having them in our test matrix creates a subtle pressure on developers to do the otherwise un-necessary work. It sort of plays on the obsessive aspect of the typical software developer personality. Robert Ramey