[EMAIL PROTECTED]"> Chris McAfee wrote:I'll go further than that - I think it is our *duty* to close the tree if checkins are causing any type of significant regressions (bustage, correctness, speed, size) rather than progress. We should only keep the tree open in such circumstances if the right engineers have traction on the problem, and closing the tree would not help them get the right fix in any faster. We don't need more quantity of code, we need better quality.
[EMAIL PROTECTED]">Part of moving forward on performance means "don't go backwards".IMO, sheriffs are well within their charter to close the tree if bloat/leak/perf numbers head south (yes, even just a little bit; unless we've consciously decided to take the hit in some tradeoff).
Peter
"Every build gets better, no exceptions."
