Test suite regressions

Tue Mar 20 13:12:00 GMT 2001

>>>>> "Jeff" == Jeff Sturm <jsturm@one-point.com> writes:

Jeff> What do you think about just delaying commits to the branch,
Jeff> maybe by a day or so? That would give at least some opportunity
Jeff> for feedback.
That's not unreasonable, but I at least have no time to try out such
patches right now. In the end I have to rely on whatever testing the
patch submitter did.
Jeff> Suppose you pick a date and call for a freeze. If there is any
Jeff> volunteer time available, post the important bugs and divide
Jeff> them. Sort of what GCC is doing as a whole.
That sounds good to me.
I've been going through the PR database and reading all the PRs.
However I quickly realized I don't have a good set of criteria to
decide what is release-critical and what is not.
In particular the problem is that there are a number of ICEs, and
places we reject valid code (even innocuous-looking valid code).
These seem like prime candidates for the "must fix" list. But it
isn't clear to me whether we have the manpower to fix them all.
Jeff> These are just ideas off the top of my head. There has been no
Jeff> talk about release methodology on this list that I can recall.
Jeff> Does anyone think a little planning would be helpful? Any other
Jeff> suggestions?
I definitely think planning is in order.
Tom