> -----Original Message----- > From: Daniel Shahaf [mailto:danie...@elego.de] > Sent: zaterdag 29 december 2012 01:00 > To: Hideki IWAMOTO > Cc: dev@subversion.apache.org; Julian Foad > Subject: Re: [PATCH] fix for diff optimization bug > > Hi Hideki, thanks for the patch. I hope Johan or Julian will have > a look (probably next year), but if you don't get a response please ping > the thread and I'll have a look.
Hi Hideki, Thanks for doing the research for this issue! I retraced your steps and applied your patch with some small tweaks in r1426752 (https://svn.apache.org/viewvc?view=revision&revision=r1426752) Thanks, Bert > > Hideki IWAMOTO wrote on Sat, Dec 29, 2012 at 08:25:07 +0900: > > Hi, Daniel. > > > > > Because I assumed using svn_diff_mem_* would reproduce issue #4133. > If > > > it doesn't, the test should be amended to use an API that will reproduce > > > the issue. > > > > The cause of issue #4133 is a bugs in the calculations of > > svn_diff__file_token_t::norm_offset. > > > > The test that you added is for different problem. (whitespaces at > > the end of line are not ignored even if option `-x -b' option is specified.) > > > > patch for issue #4133 is pattached. > > > > subversion/libsvn_diff/diff_file.c | 13 ++- > > subversion/tests/libsvn_diff/diff-diff3-test.c | 89 > ++++++++++++++++++------- > > 2 files changed, 75 insertions(+), 27 deletions(-)