Added some interesting functionality to my cygwin sandbox

Christopher Faylor
Wed Jul 2 17:36:00 GMT 2003

On Wed, Jul 02, 2003 at 07:31:35PM +0200, Corinna Vinschen wrote:
>On Wed, Jul 02, 2003 at 01:23:13PM -0400, Igor Pechtchanski wrote:
>> On Wed, 2 Jul 2003, Corinna Vinschen wrote:
>> > iscygdrive(), defined in, should be sufficient.
>> Not quite.  Suppose the user wants to create a symlink to some location on
>> his /cygdrive/c?  He should be able to do so.  IIRC, there is a difference
>> in handling EOL's when accessing files through a Win32 path and through
>> /cygdrive (you can control the binary/text mode of /cygdrive with mount
>> options).  Do we want to take this away now?
>That wouldn't happen if the implementation is correct.  Keep in mind that
>symlink target paths are *always* POSIX paths.  Now we're changing that
>by just keeping a DOS path as target.  Since we know that it's a path
>from a symlink, we can savely assume that the path was originally created
>from the corresponding cygdrive path.  So, the next step when reading that
>symlink is, revert the DOS path to a cygdrive path.  From now on, the
>usual binmode/textmode option of that /cygdrive path applies.

I think there is a disconnect here.  We aren't (or at least I'm not)
talking about making all symlinks use only windows paths.

I thought the proposal was to be able to create a special type of symlink
which contained only a DOS path.  It would be used for setting up things
which needed to always point to an arbitrary place on an arbitrary disk
not as a replacement for normal symlink behavior.  This is, I believe, why
Igor suggested a new symlink option.


More information about the Cygwin-developers mailing list