[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#6014: emacs shouldn't use C-h for "help"



In article <m0vaueD-0007qRC@i17linuxb.ists.pwr.wroc.pl>,
marekm@i17linuxb.ists.pwr.wroc.pl (Marek Michalkiewicz) writes:
> Maybe I should make it clearer: I have absolutely no problem with C-h
> if the backspace key works right (which is the case under X, or
> on terminals where the backspace key sends DEL, such as the Linux
> console).  So there is no need to change that.  But if the erase
> character is set to ^H, applications (including emacs) should
> respect that and do the right thing.

Absolutely.  I agree that Emacs should not override the user's stty'ed
erase settings.  However, I disagree that a distribution should try to
work around such problems in the application's interface.  

Fixing this is a Good Thing.  Fixing it by changing the version of the
application distributed with Debian alone is a Bad Thing, IMO.  The
right way to fix an interface issue is to get the changes into the
application itself.  Sure, it's hard in the particular case of Emacs,
but recent discussion on g.m.discuss indicates that some form of fix
will appear in 19.35 or later.  

One con of making fundamental changes to the Emacs interface is that
people who start using Emacs on a Debian system will be at sea when
faced with Emacs on the dozens of non-Debian systems that Emacs runs
on.  Another con is that such a fundamental change will mean that
dozens of user's will get answers like "use a standard Emacs" when
they pose questions on gnu.emacs.*.  A third con is that it'll anger
the maintainers of the application. (that it's RMS is irrelevant here,
I believe the tcpdump maintainers aren't very happy about the various
version of Linux tcpdump floating around; heck, I've heard people
complain of people adding code to Linux-specific utilities and never
contacting the author.)

The great thing about a distribution like Debian is that I know that
utilities know where to find things they need.  I.e., I'd fully expect
the Emacs on Debian to know where to find all the utilities a user
might want (lpr, cvs, rcs, rsh, ... the list is long).  I would never
expect the Emacs to have a different interface, though.  In fact, I'd
consider such behaviour a bug.

> Well, emacs doesn't operate well in the context of Debian (as well as
                                                             ^^^^^^^^^^
> other Unix systems) - the backspace key doesn't do the right thing when
  ^^^^^^^^^^^^^^^^^^
> running emacs in an xterm (which is part of Debian and other systems
> that come with X).

This isn't a Debian-specific problem, why provide a Debian-specific
fix? 

Also, Linux has it better than many other systems when it comes to
Emacs: it DTRT on the Linux console!  As such, it'd be kinda ironic to
change the default behaviour for the case that occurs the least often:
running inside an xterm.  The C-h problem is most acute outside of X.
Your average Debian user will _never_ encounter it.

> I don't think so.  Fixing this bug might even reduce the traffic on
> the emacs newsgroups - Debian users won't complain there about the
> backspace key not working right in xterm :-).

Um, and who'll answer questions like "the manual says I can hit C-x r
C-h and get a list of all keyseqs that start with the prefix C-x r,
but it doesn't work on my machine".  Typical responses start with
"which OS are you running?".  Answer?  "Linux".

Note that Debian will probably earn a special entry in the FAQ for
doing this.  

> <flame>
> If the upstream maintainers refuse to fix it for no good reason,
> I thing we should feel free to go ahead and fix it.  It's probably

I disagree with this from the point of view of the distribution
maintainers goals ("we'll distribute versions that are incompatible
with all other unices") and from the point of view of a heavy Emacs
user ("no good reason"? :-).

> Maybe the solution to keep everyone happy would be to ask the user
> in emacs postinst, whether he wants the bug fixed?

a) I would not call it a bug.  It is petty at best, and offensive to
   the program's maintainers at worst.
b) This is not a big issue for Linux based systems.
c) Debian shouldn't be meddling with the interface.

After all that verbiage, I admit I like this suggestion.

-Sudish


--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-REQUEST@lists.debian.org . Trouble? e-mail to Bruce@Pixar.com


Reply to: