No such file or directory

Larry Hall (RFK Partners, Inc)
Mon Mar 27 07:25:00 GMT 2000

At 12:05 PM 3/26/00, you wrote:
>On Sat, Mar 25, 2000 at 11:07:27PM -0500, Larry Hall (RFK Partners, Inc) wrote:
> >At 10:48 PM 3/25/00, Chris Faylor wrote:
> >>On Sat, Mar 25, 2000 at 10:40:13PM -0500, Larry Hall (RFK Partners, Inc) wrote:
> >> >Looks to me like your big problem is the install program which assumes the 
> >> >UNIX convention of executables without extensions (.exe).  It looks for
> >> >remind and not remind.exe, which is what exists.
> >>
> >>I wonder if it would really be a big deal if cygwin, by default, found a
> >>file "foo.exe" if there was no existing file "foo".
> >>
> >>We keep running into this problem and I wonder if implementing this in
> >>cygwin would solve more problems than it causes.
> >
> >Its been a while but isn't there already something in there that does the
> >reverse for executables in general?  I should go looking.  My guess is what
> >I'm remembering is a related but different topic.
>The stat() function call will locate a file with no extension and one
>with a .exe extension.  Obviously, the spawn() and exec() calls do
>something similar.
>The problem is that if you do a stat("ls", &st) it will succeed but
>open("ls", whatever) will fail.

Ah, OK, stat() would've been what I was thinking of.  I've spent some time
there!;-)  I didn't realize there was an difference between what stat() did
and what open() did in this regard.

> >But I agree with you.  This is a common "problem" of the
> >configure/make/ install sequence of GNU packages on Cygwin and,
> >although its not a big deal to work with once you understand it, it
> >does generate list traffic regularly.  I guess I'm wondering why the
> >alternate install script solution that has been "floated" in the past
> >is not something Cygwin uses now.  I admit I never looked that closely
> >at it.  If it does the right job, the thing that I think is good about
> >it is that it targets the problem in the context of installs, which
> >seems to be where the problem lies...
>install.exe in the next version of cygwin will be ".exe" aware.
>It hasn't been done before now because, ah, we haven't had an official
>net release in a long time.  The Cygwin CD version does use this version
>of install, however.

Right.  I figured the install issue was mostly a latency thing.  Anyway, in
my opinion, at least the category of problem I see popping up in the list
allot would be resolved by this change.  This seems adequate to me (although
once its in place for the populace in general, I might decide otherwise!;-))

Thanks for the info/clarification.


Want to unsubscribe from this list?
Send a message to

More information about the Cygwin mailing list