21 Nov
2010
21 Nov
'10
1:23 p.m.
On Nov 21, 2010, at 11:51 AM, Vincent Massol wrote:
On Nov 21, 2010, at 11:50 AM, Vincent Massol wrote:
Not sure this email went through, resending.
On Nov 21, 2010, at 9:12 AM, Caleb James DeLisle wrote:
* Bayesian filter. When we see false positives we teach the filter. It works quite well
for spam it should work for this too. How do you know that these are false positives?
Some ideas:
* We could have a list of known false positive messages:
** "Premature end of file"
** "Failed to start XWiki in [120] seconds"
** "Internal Server Error Response"
** etc
* We can also check if several tests have the same output. Below several have the same
500 error.
Thanks
-Vincent
[snip]