9 Apr
2008
9 Apr
'08
3:14 p.m.
Peter Dimov wrote:
John Maddock:
Peter Dimov wrote:
The tests are indeed too permissive... but the real reason is that config_test is always run with <threading>multi, so the default configuration is never tested. Is this correct? Most of our other tests do not use <threading>multi, so it seems reasonable to run config_test in this mode as well (in addition to the existing <threading>multi run).
Ah, I'll add another build variant for that test.
Can you please add it for config_info as well? Oddly, it seems that currently it's being run without <threading>multi.
Done, John.