Justin Erenkrantz <[EMAIL PROTECTED]> writes: > --On Monday, March 10, 2003 12:21 AM -0600 [EMAIL PROTECTED] wrote: > > > call. The 'dirstruct' member of that structure was particularly > > interesting to me, and looked like this after each call (the paths > > after the '###' below are the current dirent just read by > > svn_io_dir_read()): > > Can you jump into apr_dir_read and check the return values from > readdir_r? I'd imagine that we should be hitting the !ret && > thedir->entry != retent case on line 187 - that should be returning > APR_ENOENT. > > Is that the case? -- justin
First, let me say this: http://www.cs.helsinki.fi/linux/linux-kernel/2001-42/0064.html Then, let me say that yes, this bug only happens for me when I'm running the tests on tmpfs. Finally, let me crawl into a deep dark hole and lie there a while. (Karl ask me Friday afternoon if I had tested on both tmpfs and on the regular filesystem. I was confident that I had, but apparently messed up somewhere.) Oh, Release Manager -- can you put the 5202 change back into the release, pretty please?
