On Tue, Sep 9, 2008 at 12:55 AM, Stephen J. Turnbull
<turnbull(a)sk.tsukuba.ac.jp> wrote:
Lars Nilsson writes:
> When opening files on Windows, XEmacs does not maintain the filename
> case for existing files.
While your request is plausible, I don't see that XEmacs necessarily
should preserve case if neither Windows nor cl.exe tries to, and if a
separate directory listing is required to get the file name, doing the
check is potentially expensive (the file may be on a network server).
What I was interested in was similar behaviour to tab-completion of
the filename when manually opening a file with C-x C-f for instance,
which will present me with the proper capitalization regardless of
what I started typing, but in a different context (in this case,
next-error).
I certainly considered the possible performance impact for large
collections of files in one directory, slow access (network, or
otherwise). In my particular case, given the distribution of files,
etc, it did not seem like it would have a big impact, and I haven't
experienced any big problems. Of course, I'm not stressing XEmacs with
load operations, like compiling a lot of .el files, or whatever else
might involve a lot of calls to find-file (if not relevant to
compile-file and friends, have never looked at how deep the
utilization of find-file goes).
> I would like to stress that I would feel more comfortable if
XEmacs
> maintained the case for existing files (without me poking at stuff
> I don't necessarily understand) regardless of what case was
> specified in the request.
"Regardless" obviously is a non-starter; what if the user really wants
to change the case of the file name? So it becomes a question of
figuring out when XEmacs should implicitly change the command it
receives.
The user is only me. I don't pretend to speak for anyone else. I
wanted it, so I wrote a solution that works for me. I was just curious
if there was another, better way, or perhaps even something of
interest to others.
I can see two possibilities for existing tools or settings that
would
work in your context. The first is that PCL-CVS may "do the right
thing".
Being on Windows, I'm using a Windows Explorer Shell integration for
CVS activities. Just my preference at the moment, of course.
Second, you could try setting `backup-by-copying' to t, which
might
leave the directory entry of the file alone. You might also need to
disable auto-save or set `auto-save-visited-file-name' t.
Having tested it, it would seem that setting backup-by-copying and
auto-save-visited-file-name to t does preserve the filename case, just
the way I'd like it. I have not tried them separately, but it seems to
render my own hack irrelevant. Thank you very much.
Regards,
Lars Nilsson
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta