From: Nguyen Thai Ngoc Duy Date: Mon, 25 May 2009 06:34:52 GMT Subject: Re: [PATCH v0 3/3] Build in git-rebase.sh Message-ID: In-Reply-To: <4A1A37AD.4080309@viscovery.net> On Mon, May 25, 2009 at 4:16 PM, Johannes Sixt wrote: > Nguyen Thai Ngoc Duy schrieb: >> On Sat, May 23, 2009 at 07:26:03PM +1000, Nguyen Thai Ngoc Duy wrote: >>> On Fri, May 22, 2009 at 05:30:31PM +1000, Nguyen Thai Ngoc Duy wrote: >>>> 2009/5/22 Johannes Sixt : >>>>> Nguyễn Thái Ngọc Duy schrieb: >>>>>> Signed-off-by: Nguyễn Thái Ngọc Duy >>>>> Is it possible for you to test this series on Windows? Many rebase tests >>>>> fail, but I haven't investigated why. >>>> I'll try it this weekend. >>> This patch makes t3*rebase*.sh pass for me except t3412 (more exactly >>> t3412.8). That test failed even with git-rebase.sh. Hmm... Anyway >>> could you try again to see what tests still fail? >> >> Someone with better Windows knowledge than me should explain how this works. Windows' >> snprintf() just cuts out the last character in this statement: >> >> snprintf(buf, 8, "--%s", "onto"); // result: '--ont', expected: '--onto' > > This doesn't happen for me: neither with Windows's original snprintf nor > with the version from compat/. Are you using the latest msysgit > environment to compile, i.e. gcc 4.4? There was a change regarding > SNPRINTF_SIZE_CORR; perhaps that's the culprit? It could be. I used git.git's next branch, not msys4git or mingw.git repositories. > I don't undertand what this patch does, anyway. Where is the detail that I > am missing? The first vsnprintf() in "onto" case above did not produce " --onto" as it should have been. I tried to recover by filling "str" with the second vsnprintf() in the loop. >> All rebase tests now pass for me on Windows (Vista something, I have >> yet to find where it hides its "uname" command) > > They also pass for me with your earlier fix-up patch, but with or without > this patch to compat/snprintf.c. Good to know. -- Duy