On Thu, Nov 23, 2017 at 9:25 PM, Elijah Newren <new...@gmail.com> wrote: > On Thu, Nov 23, 2017 at 2:28 PM, Elijah Newren <new...@gmail.com> wrote: >> On Thu, Nov 23, 2017 at 3:52 AM, Adam Dinwoodie <a...@dinwoodie.org> wrote: >>> On Tuesday 21 November 2017 at 12:00 am -0800, Elijah Newren wrote: >>>> <snip> >>>> >>>> merge-recursive.c | 1243 +++++++++++- >>>> merge-recursive.h | 17 + >>>> t/t3501-revert-cherry-pick.sh | 5 +- >>>> t/t6043-merge-rename-directories.sh | 3821 >>>> +++++++++++++++++++++++++++++++++++ >>>> t/t7607-merge-overwrite.sh | 7 +- >>>> unpack-trees.c | 4 +- >>>> unpack-trees.h | 4 + >>>> 7 files changed, 4985 insertions(+), 116 deletions(-) >>>> create mode 100755 t/t6043-merge-rename-directories.sh >>> >>> The new t6043.44 introduced in this branch is failing on my Cygwin >>> system. I can't immeditely see what's causing the failure, but I've >>> copied the relevant verbose + shell tracing output below in the hope it >>> makes more sense to you: >> >> Thanks for reporting. Unfortunately, I have been unable to locate or >> create a cygwin system on which to replicate the testing. Valgrind is > > Nevermind; found a system that allowed me to reproduce the problem, > even if it far more wrangling than I'd like. I believe I have a > one-line fix, but I'm worried that attempting to fully explain it will > result in a novel-length commit message.
This issue is addressed by the commit at https://public-inbox.org/git/20171124195901.2581-1-new...@gmail.com/ . I opted to make it a different series, because from my view it looks like a separate latent bug in the ignore_case handling that was just unearthed by a combination of events that included my series. It's conceivable it could have eventually been triggered some other way. However, that commit cleanly cherry picks to any of maint, master, next, or pu, so If Junio just wants to just include that commit on the top of the en/rename-directory-detection series, that's fine too. Without that commit, I get the same failure you did on cygwin. With that commit, I get all tests passing on cygwin for pu as of yesterday. (Well, all tests that ran; I didn't have svn or apache or p4 or various other things installed. Also, there were some unexpected passing TODOs, but I've already seen others discussing those exact testcases on the list elsewhere.) Hope that helps, Elijah
