On Tuesday, January 21, 2003, at 07:43 PM, Beman Dawes wrote:
At 12:23 PM 1/21/2003, Matthias Troyer wrote:Yes, you are correct and I am tracking down the problem. It seems that there are some problems starting the regression tests: bjam runs for 10 minutes, without writing anything into the log file, and without actually doing any compilations. Since nothing is actually done, of course these directories don't exist.
>I could get jam to compile and could start the regression tests with a
>couple of patches to Makefiles and scripts. Now I am stuck, since for
>each library the regression tests abort with an error message such as:
>
>*** Error: std::runtime_error: boost::filesystem::directory_iterator
>constructor: "/u/ph/troyer/boost/libs/type_traits/test/bin": No such
>file or directory
>
>Should these directories be created automatically, or should they
>already be present? For now I can just create all of them by hand, but
>I would like to find the origin of this problem. Can anyone help me?
That message looks like it is coming from process_jam_log or compiler_status. If so, it is caused by bjam putting outputs in one location but then the post-bjam processing steps looking for them in another location.
I already had to patch this file in a couple of places since the Cray 'test' command does not work as on other machines.
If that doesn't give you enough information to fix the problem, please post the actual commands you are using, including the initial cd which sets the current directory. Also, you might want to look at boost-root/tools/regression/run_tests.sh if you haven't done so already.
I guess the best way to proceed is if I run the regression tests on a machine where they work, to see what the output should be, and then to see where the Cray differs.
Matthias
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
