Vladimir Prus wrote:
Martin, your last run has a bunch of failures caused by errors in Boost.Test as shown here:
Probably, somebody checked in something that broke Boost.Test on this specific compiler.
The recent test cycle took rather long. That usually indicates a change to a core component. ("core" includes "Boost.Test", in this case).
Other ideas are:
1. There's some hardware issue, like flaky RAM. But the error is the same everywhere.
The "but" part doesn't contradict the flaky-RAM assumption, actually.
2. There was some cvs conflict on update, but only one toolset if affected.
That's unlikely.
Can you probably do this:
1. Manually try building Boost.Test for gcc-3.3.6 to verify it's not a sporadic error.
I reran the tests for the all toolsets right after I saw the results. Results should show up with the next table update.
2. Try to backdate CVS a few days and see if that helps.
That's a bad idea. Rerunning the tests under that condition would take several days. Switching back to the head revision of the RC branch would then take another couple of days. Regards, m Send instant messages to your online friends http://au.messenger.yahoo.com