Using bugzilla to track XFAILs
Giovanni Bajo
giovannibajo@libero.it
Fri Feb 13 14:28:00 GMT 2004
Hello,
it looks like we currently don't have any way to track regressions which are
XFAILed in the testsuite. Those regressions are not less important than the
other regressions which get submitted by the users, but still it doesn't seem
that developers usually care that much about them, and they are not used, for
instance, to set goals for a release. It already happened to me two times to
fix bugs after Bugzilla reports and find out that there was already a testcase
for this that was XFAILed; and I submitted only a few patches till now. It
seems silly that we have to wait for users to rediscover regressions to get
them fixed.
There must be a better way to not forget about them. I propose the introduce a
new policy: for each regression which has been xfailed in the testsuite, there
ought be a bugzilla bug referencing it. We already have a keyword called
'xfail', but it is currently used only by the tree-ssa branch.
Does this seem reasonable? If so, I volunteer to handle the details and prepare
the documents (HTML?) to illustrate this policy to the developers.
Giovanni Bajo
More information about the Gcc
mailing list