"John Maddock" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED] > The following may well be boost.build problems, but: > > There are tests showing up in the latest results that have been > removed/renamed in the latest jamfiles. > There are tests listed as lib build failures (Borland/Metrowerks on Win32) Looking at http://boost.sourceforge.net/regression-logs/cs-win32_metacomm/developer_result_page.html#regex and http://boost.sourceforge.net/regression-logs/cs-win32_metacomm-links.html#libboost_regex-cwpro8
> even though the lib most certainly does build (the other tests for these I don't see that the library builds OK for cwpro8. Strangely, Beman's test results are fine. > compilers are OK, and there is nothing in the logs for the lib build, and in > the case of Borland the tests build OK for me). > There are tests shown as failures because bjam has not correctly forced a > rebuild of the lib (specific to Linux - I can reproduce this - only deleting > the old .a and .o files clears the problem). > So... can all the regression testers please delete the libs/regex/*/bin > directories to clear these? We always to "clean" builds, we delete everything, do fresh checkout from CVS and rebuild everything except bjam. The reasons for it are: 1. It seems to me that bjam doesn't seem to track all dependencies (i.e. the dependencies on toolsets) 2. We wanted to make sure that what we are in complete sync with CVS. Misha Bergal MetaCommunications Engineering _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost