bug#3761: 23.1.50; TAB file name completion skips valid entry

[Kevin Rodgers]

Lennart Borgman wrote:
> On Sun, Aug 16, 2009 at 6:55 PM, Eli Zaretskii<eliz@gnu.org> wrote:
> > > Date: Sun, 16 Aug 2009 12:59:22 +0200
> > > From: Lennart Borgman <lennart.borgman@gmail.com>
> > > Cc: 3761@emacsbugs.donarmstrong.com
> > >
> > > On Sun, Aug 16, 2009 at 7:31 AM, Chong Yidong<cyd@stupidchicken.com> wrote:
> > > > > I have two directories:
> > > > >
> > > > >   ~/ecb-cvs/
> > > > >   ~/ecb-cvs-old/
> > > > >
> > > > > Doing
> > > > >
> > > > >   C-x C-f ~/ecb TAB
> > > > >
> > > > > gives ~/ecb-cvs-old/ which is not what I want.
> > > > I don't see this problem.  With this recipe, I get ~/ecb-cvs
> > > >
> > > > Are you sure?
> > >
> > > Yes, I just checked again with a checkout from 2009-08-01. (Starting
> > > from "emacs -Q" of course.)
> > >
> > > It could perhaps be w32 specific. Could someone more please check?
> > Hint: see the value of completion-ignored-extensions.
>
>
> Hm, I see. ;-)
>
>
> > Okay to close this bug as a feature?
>
>
> Maybe, I am not sure. Is this the intended behaviour? Should all file
> names ending in "CVS/" be ignored?

It doesn't seem right to me either.  What happens if you change "CVS/"
in completion-ignored-extensions to "/CVS/"?

> Does CVS use any other file names
> beside just "CVS"?

I think CVS is hard-wired to use "CVS", but it's been a while since
I used it.

> Why is not the hidden property on the file name in
> this case used instead? (At least on w32 the CVS directory has a
> hidden property.)

--
Kevin Rodgers
Denver, Colorado, USA