Bert, On Thu, Jul 21, 2011 at 2:04 PM, Bert Huijben <b...@qqmail.nl> wrote: > > >> -----Original Message----- >> From: l...@apache.org [mailto:l...@apache.org] >> Sent: donderdag 21 juli 2011 13:08 >> To: comm...@subversion.apache.org >> Subject: svn commit: r1149116 - /subversion/branches/1.7.x/STATUS >> >> Author: lgo >> Date: Thu Jul 21 11:07:44 2011 >> New Revision: 1149116 >> >> URL: http://svn.apache.org/viewvc?rev=1149116&view=rev >> Log: >> * STATUS: Reviewed and tested issue3888 branch: +1 -> approved. > > Did you review the branch (two trivial changes) or the behavior change? :) >
I know how this code should behave, as I: 1. already had a first try on solving this issue: http://svn.apache.org/viewvc?view=revision&revision=1081141 2. Discussed with Ivan some of the drawbacks of my implementation: http://svn.haxx.se/dev/archive-2011-03/0440.shtml 3. Looked at and reviewed some of Greg's changes: http://svn.haxx.se/dev/archive-2011-06/0503.shtml This morning I spent two hours reviewing the implementation, and testing with large checkouts and updates that the mechanism is enabled and solves the memory leak problem, both with trunk and 1.7.x on Mac OS X. For me that gave me enough confidence to add my +1 to the backport votes. Does that mean this code is bug-free? Probably not, but there is nu such guarantee for any of the other backported changes either. Since you are openly doubting my integrity in testing and approving changes for backport, I propose you review the code changes yourself and add your own +1 or -1 vote. Lieven
