From: Doug Gregor (dgregor_at_[hidden])
Date: 2007-12-13 11:51:49
I've turned off the report generation for noon today, and will try to
get improved reports out later today. Some comments below...
On Dec 13, 2007, at 11:33 AM, Beman Dawes wrote:
> Douglas Gregor wrote:
>> Boost regression test failures
>> Report time: 2007-12-12T11:20:18Z
>> This report lists all regression test failures on release platforms.
> Doug, please change this to "high priority" platforms.
>> Detailed report:
>> The following platforms have a large number of failures:
>> 3311 failures in 71 libraries (118 are from non-broken platforms)
> One of the Sandia gcc-4.2.1 machines ran out of disk space, and
> there is some other platform-wide problem causing Sandia intel-
> linux-9.0 failures.
> The failures on these two platforms completely obscure the real
> failures we need to be worrying about. Since you are already
> recognizing that there is something really wrong, could you not
> report the detail for these platforms? Anyone who wants to see the
> details can go to the regression reports on the web.
Sure, I can do that. I'll put the total number of failures after the
platform name, and then omit all of that information in the detailed
report below (including the reports sent to Boost developers).
Actually, this is one of those cases where it would be great to start
really using libs/platform_maintainers.txt. We can e-mail the
platform maintainer when there is a large number of failures on that
platform, so s/he can look into it.
>> algorithm/minmax (0 of 4 failures are from non-broken platforms)
> Maybe a legend at the top that explains what this means? Or am I
> the only one who wonders what a "non-broken platform" is?
I'll rephrase this in the context of the other changes.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk