On 9/5/2011 3:08 AM, Gregg L. Smith wrote: > > regressions suck, but the killer is silenced as far as I can tell at this > point, the > script fails it's internal test, no need to rush and gain more regressions > hurrying to fix > the regressions from hurrying to silence the killer. My meager view anyway :)
I agree except that we already rushed .20 and suggest somehow that it is the "best available version of httpd", which is nonsense ;-) If the rate of new issue reports has stalled, and Jim's willing to tag Wednesday, then I'd agree we should do it. Right now we are asking people to test a broken flavor against clients, so if we ship with a protocol-valid flavor we'll get more useful defect reports which either httpd or the client authors can choose to address, depending on the actual reasons for breakage. Right now, we know clients should be broken, so asking anyone to report to us that 2.2.20 breaks some obscure clients is almost useless to us and to the authors of those clients.
