On Dec 4, 2005, at 3:44 PM, Daryle Walker wrote:
On 12/4/05 12:14 PM, "Douglas Gregor" <doug.gregor@gmail.com> wrote:
On Dec 4, 2005, at 12:01 PM, Douglas Gregor wrote:
1015 failures in 56 libraries:
[most failures are from one compiler; most libraries only have problem with that one compiler; said compiler's appearance is a surprise and it'll usually go away in the next cycle]
Rene is looking into these... let's hope it's some kind of simple configuration problem...
Maybe we should add something to a script that if a compiler shows up in most tests, especially if it's the sole problem compiler for many tests across many libraries. The message could say that "compiler X by operator Y seems to be broken." (My criteria may be too vague for a computer.)
Yes, this is a good idea. The script that e-mails these reports is tools/regression/xsl_reports/email_maintainers.py and should be pretty easy to hack, if you're interested in taking a shot at adding this functionality. Doug