Re: cygwin_conv_path sometimes removes trailing slash

[Ray Donnelly <mingw.android@gmail.com>]

On Mon, Feb 8, 2016 at 6:13 PM, Corinna Vinschen
<corinna-cygwin@cygwin.com> wrote:
> Hi Ray,
>

Hi Corinna,

Sorry for the delay, I've been snowed under with day-job things lately.

> On Jan 31 22:17, Ray Donnelly wrote:
>> On Sun, Jan 31, 2016 at 11:38 AM, Jan Nijtmans <jan.nijtmans@gmail.com> wrote:
>> > 2016-01-30 4:21 GMT+01:00 Ken Brown:
>> >> I'm using cygwin_conv_path to convert Win32 paths to POSIX paths, and I'm
>> >> puzzled by the conversion
>> >>
>> >>   d:/ --> /cygdrive/d
>> >>
>> >> without the trailing slash.  By contrast, we have
>> >>
>> >>   d:/foo/ --> /cygdrive/d/foo/
>> >
>> > This came up before:
>> >    <https://cygwin.com/ml/cygwin/2015-10/msg00048.html>
>> > I had to workaround it in tkimg.
>> >
>> > It's easy to workaround in your program, but - indeed -
>> > that shouldn't be necessary. I too would expect that
>> > when the input contains a slash, the output should too.
>> >
>>
>> I've attached the patch that we use on MSYS2 for this. We found
>> ourselves patching projects all over the open-source landscape
>> otherwise
>
> How many projects require exposure to the conversion from POSIX to Win32
> paths?  This is something which should be used in applications only very
> rarely.  Usually they just feed the POSIX path into functions like open,
> etc., and that's it.
>
>> so fixing it once and for all seems sensible. Note however,
>> a full fix also requires another patch that I posted two years ago [1]
>> that Corinna chose not to apply. In the two years since, we've yet to
>> find any issue with that patch. I will rebase it shortly in-case
>> there's any interest in reconsidering it.
>>
>> [1] https://cygwin.com/ml/cygwin-patches/2014-q1/msg00010.html
>
> Your patches don't work for Ken's case since path_conv::check is never
> called for CCP_WIN_A_TO_POSIX/CCP_WIN_W_TO_POSIX conversion.  I have a
> patch in the loop to fix that.

Great!

>
> I'm not quite sure what your first patch is supposed to accomplish in
> the POSIX->WIN case.  tail == path_copy + 1 means the incoming path is
> "/".  Your patch will result in changing the conversion of
>
>   /  -->  C:\cygwin
>
> to
>
>   /  -->  C:\cygwin\
>
> Is that really desired?  What especially bugs me is that this occurs in
> the very core of the path conversion which isn't only used in places
> exposed to user space.

It is what we needed on MSYS2. When building a "native" GCC, we'd pass
e.g. --sysroot=/ and that'd get expanded to "C:/cygwin" and have say a
sys include dir of "sys/include" appended making
C:/cygwinsys/include". I can't guarantee that *nothing* else will
break because of this change except to offer that anecdotally no msys2
package (i.e. those packages in MSYS2 that are purely POSIX) build or
execution failure was ever determined to be caused by making this
change.

Best regards,

Ray.

>
>
> Thanks,
> Corinna
>
> --
> Corinna Vinschen                  Please, send mails regarding Cygwin to
> Cygwin Maintainer                 cygwin AT cygwin DOT com
> Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple