This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [ANNOUNCEMENT] Updated: cygwin-1.5.19-1


On Wed, Jan 18, 2006 at 07:10:27PM +0100, Corinna Vinschen wrote:
>On Jan 18 12:27, Sam Steingold wrote:
>> > * Christopher Faylor <pts-ab-crefbany-ercyl-cyrnfr@pltjva.pbz> [2006-01-16 22:36:19 -0500]:
>> >
>> > corinna: Emulate linux version of realpath
>> 
>> now when I have a file "foo.exe",
>> realpath("foo") returns "foo" instead of "foo.exe"
>> also:
>> $ ls -l /proc/self/exe
>> 0 lrwxrwxrwx 1 sds mkgroup-l-d 0 Jan 18  2006 /proc/self/exe -> /usr/bin/ls*
>> 
>> instead of ls.exe, so /proc/self/exe cannot be open() for input.
>> 
>> so, now I must add the "exe" suffix myself?!
>
>Funny that you mention it.  The realpath change to "emulate linux"
>has nothing to do with what you're talking about.  The "emulate linux"
>change is something about how to behave if the second argument is a
>NULL pointer.  Linux returns a malloced buffer in this case, which
>Cygwin didn't so far.
>
>Now, talking about the .exe suffix... the realpath change is this:
>
>  Before: realpath("foo", ...) = NULL (ENOENT)
>  After:  realpath("foo", ...) = "foo".
>
>So, instead of returning ENOENT it returns a path that the file exists
>which is equivalent to how stat behaves, which it wasn't before.
>
>The only problem is that this change was already applied on 2005-01-25,
>which means this is in Cygwin since release 1.5.13 (2005-03-01).
>
>As for removing the .exe from /proc/$PID/exe(name), AFAICS this hasn't
>changed since at least 2004-01-01, which was the date when I gave up
>looking for a change in the code.
>
>So, well, I don't see why you "now" would have to change something
>exactly in relation to this.

FWIW, I have noticed inconsistent behavior between *all* of the new
functions which were added to cygwin 1.5.19.  Older versions of cygwin,
which lacked the functions, behaved quite differently from newer
versions of cygwin which now have the functions.

Someone should look into this, as it is obviously a regression.

cgf

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


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]