Re: small Eshell docfix

Re: small Eshell docfix

[Richard Stallman]

      (defcustom eshell-explicit-command-char ?*
	"*If this char occurs before a command name, call it externally.
      That is, although vi may be an alias, *vi will always call the
    ! external version.  UNIX users may prefer this variable to be \\."
	:type 'character
	:group 'eshell-ext)

When it says "UNIX users", who does that mean?
Does it mean users of Solaris and BSD as opposed to GNU/Linux?
Why would they be different?

Or does it intnded to mean users of GNU/Linux, Solaris, and BSD as
opposed to Windows?  Whatever's best for GNU/Linux should be the
default.

Re: small Eshell docfix

[John Wiegley]

Richard Stallman <rms@gnu.org> writes:

>       (defcustom eshell-explicit-command-char ?*
> 	"*If this char occurs before a command name, call it externally.
>       That is, although vi may be an alias, *vi will always call the
>     ! external version.  UNIX users may prefer this variable to be \\."
> 	:type 'character
> 	:group 'eshell-ext)
>
> When it says "UNIX users", who does that mean?  Does it mean users
> of Solaris and BSD as opposed to GNU/Linux?  Why would they be
> different?

Actually, I think everybody would prefer it to be *.  The only reason
I said that in the comment is that ksh uses "\word" to override any
alias defined for "word".  Feel free to remove that line from the
documentation.

John

Re: small Eshell docfix

[Stefan Monnier]

> Actually, I think everybody would prefer it to be *.  The only reason
> I said that in the comment is that ksh uses "\word" to override any
> alias defined for "word".  Feel free to remove that line from the

Not only ksh.  Pretty much all shells from either the bourne or csh family
use that convention it seems.


        Stefan

Re: small Eshell docfix

[Kim F. Storm]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> > Actually, I think everybody would prefer it to be *.  The only reason
> > I said that in the comment is that ksh uses "\word" to override any
> > alias defined for "word".  Feel free to remove that line from the
> 
> Not only ksh.  Pretty much all shells from either the bourne or csh family
> use that convention it seems.

I actually mis-read the original doc strings several times before getting
the meaning ...

 external version.  UNIX users may prefer this variable to be \."
 external version.  UNIX users may prefer this variable to be \\."

I read it as .. "to be period"  rather than "to be backslash" ...

A better wording (that also avoids mentioning UNIX) may be:

 external version.  Users of bash and similar shells may 
 prefer to use \\ here.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: small Eshell docfix

[Richard Stallman]

    Not only ksh.  Pretty much all shells from either the bourne or csh family
    use that convention it seems.

In that case, why isn't \ the default for this variable?

Re: small Eshell docfix

[John Wiegley]

Richard Stallman <rms@gnu.org> writes:

> In that case, why isn't \ the default for this variable?

Because this would conflict with Windows, where "\vi" means to run the
vi.exe that exists in the root directory.

John

Re: small Eshell docfix

[Richard Stallman]

    > In that case, why isn't \ the default for this variable?

    Because this would conflict with Windows, where "\vi" means to run the
    vi.exe that exists in the root directory.

The idea that an Emacs feature should be designed to avoid "conflict
with Windows" gives Windows more importance than it deserves.

GNU Emacs is a part of the GNU system and its main purpose is to make
the GNU system better.  It is ok for GNU packages to support non-free
systems as well, as long as they don't get in the way of the main
purpose.

If we start rejecting decisions that make sense on GNU/Linux because
they "conflict" with Windows, Windows is getting in the way.

Re: small Eshell docfix

[Oliver Scholz]

Richard Stallman <rms@gnu.org> writes:

>     > In that case, why isn't \ the default for this variable?
>
>     Because this would conflict with Windows, where "\vi" means to run the
>     vi.exe that exists in the root directory.
>
> The idea that an Emacs feature should be designed to avoid "conflict
> with Windows" gives Windows more importance than it deserves.
>
> GNU Emacs is a part of the GNU system and its main purpose is to make
> the GNU system better.  It is ok for GNU packages to support non-free
> systems as well, as long as they don't get in the way of the main
> purpose.
[...]

I'd like to point out that in my view Emacs' relation to Free Software
is deeper than that it is designed to make the GNU system better.

For me at least it was Emacs that let me understand the values of Free
Software as opposed to merely "open source". When I started to use
GNU/Linux it was just because the open-source hype and because I
wanted to get away from MS Windows 98. That in theory I could review
the sources of packages that I used and that I was entitled to change
them had no meaning to me at that time: I wasn't able to program.
Sure, even at that time I could say to myself: well, if it were
necessary, then I /could/ learn how to program and change the sources
... but why should I? Until you actually /experience/ the benefits of
having the sources available, such possibilities do not mean much to
you.

But when I gained a better understanding of Emacs, this point of view
changed. The Emacs community habitually publishes Emacs Lisp snippets
both in Emacs related newsgroups and mailing lists and on the web.
"You have a problem, you want a feature? I took a glance at the
sources -- here are a few lines of code that solve it." Even if you
are not a programmer, this is a way to actually experience what Free
Software means: not just a claim to technical superiority ("open
source"), but the freedom to look at and change. And this was what
made me learn Emacs Lisp.

So I say: Emacs is the best advertising medium, the best promotion of
the idea of Free Software.  So, in my point of view there is value in
itself if it runs smoothly on non-free operating systems like MS
Windows or MacOS. It is, so to speak, a foot in the door.

    Oliver
-- 
Oliver Scholz               16 Fructidor an 212 de la Révolution
Ostendstr. 61               Liberté, Egalité, Fraternité!
60314 Frankfurt a. M.       

Re: small Eshell docfix

[Richard Stallman]

    So I say: Emacs is the best advertising medium, the best promotion of
    the idea of Free Software.  So, in my point of view there is value in
    itself if it runs smoothly on non-free operating systems like MS
    Windows or MacOS. It is, so to speak, a foot in the door.

All else being equal, it is useful that Emacs runs on these systems.
But that's not the issue at hand.  The issue at hand is whether to
design Emacs features to work better in the context of the GNU system
or to design them to work better in the context of Windows, in cases
where those two goals conflict.

Re: small Eshell docfix

[Emilio Lopes]

Richard Stallman writes:

>       (defcustom eshell-explicit-command-char ?*
> 	"*If this char occurs before a command name, call it externally.
>       That is, although vi may be an alias, *vi will always call the
>     ! external version.  UNIX users may prefer this variable to be \\."
> 	:type 'character :group 'eshell-ext)

> When it says "UNIX users", who does that mean?  Does it
> mean users of Solaris and BSD as opposed to GNU/Linux?  Or
> does it intnded to mean users of GNU/Linux, Solaris, and
> BSD as opposed to Windows?

The later, I would say.

Do you think "Users of most other shells might prefer..." or
something like that would be better?

> Whatever's best for GNU/Linux should be the default.

That would probable be `?\\', but `?*' is already the
default for sometime now.

Re: small Eshell docfix

[Richard Stallman]

    That would probable be `?\\', but `?*' is already the
    default for sometime now.

It seems \ is a better default, makes the GNU system more coherent
overall.  I think people will get used to the change.

Re: small Eshell docfix

[John Wiegley]

Richard Stallman <rms@gnu.org> writes:

> It seems \ is a better default, makes the GNU system more coherent
> overall.  I think people will get used to the change.

I have always thought one of Emacs' best selling points was that it
provides a consistent environment across nearly all operating systems.
That, at least, was a huge draw for me: I can start up Emacs on
Windows, without modification, and lose none of its functionality.

The change you are proposing takes the emphasis away from Emacs, and
puts it on GNU/Linux.  Since we are talking about Emacs, why all the
hoopla about GNU/Linux?

I chose the defaults I did so that Eshell would behave similarly on
all platforms.  I ask that you please not change such an excellent
design philosophy now.

John

Re: small Eshell docfix

[Richard Stallman]

    The change you are proposing takes the emphasis away from Emacs, and
    puts it on GNU/Linux.  Since we are talking about Emacs, why all the
    hoopla about GNU/Linux?

GNU Emacs is a part of the GNU system, and its main purpose is to make
the GNU system better.  That is why I wrote it.  It would be a mistake
to get distracted by other systems.

>From Information For Maintainers of GNU Software:

    The most important platforms for a GNU package to support are GNU and
    GNU/Linux.  Developing the GNU operating system is the whole point of
    the GNU Project; a GNU package exists to make the whole GNU system more
    powerful.  So please keep that goal in mind and let it shape your work.
    For instance, every new feature you add should work on GNU, and
    GNU/Linux if possible too.  If a new feature only runs on GNU and
    GNU/Linux, it could still be acceptable.  However, a feature that runs
    only on other systems and not on GNU or GNU/Linux makes no sense in a
    GNU package.

    ...

    It is important to test the program personally on GNU or GNU/Linux,
    because these are the most important platforms for a GNU package.  If
    you don't have access to one of these platforms, please ask
    @email{maintainers@@gnu.org} to help you out.

    Supporting other platforms is optional---we do it when that seems like a
    good idea, but we don't consider it obligatory.

To give Windows and GNU equal consideration in Emacs development is
out of the question.  Any argument which leads to that conclusion is a
reductio ad absurdam.

Re: small Eshell docfix

[Emilio Lopes]

Richard Stallman writes:

[ refering to `eshell-explicit-command-char' ]

>     That would probable be `?\\', but `?*' is already the
>     default for sometime now.

> It seems \ is a better default, makes the GNU system more
> coherent overall.  I think people will get used to the
> change.

Here is a patch.  This change should probably be announced in the NEWS
file.  Something like

** The default value of `eshell-explicit-command-char' was
changed to `\' to be consistent with the rest of the GNU system.


2004-09-01  Emilio C. Lopes  <eclig@gmx.net>

	* eshell/esh-ext.el (eshell-explicit-command-char): Changed
	default value to ?\\.  Modified docstring accordingly.


*** lisp/eshell/esh-ext.el.~1.15.~      Mon Jan  5 13:27:42 2004
--- lisp/eshell/esh-ext.el      Wed Sep  1 22:06:27 2004
***************
*** 148,157 ****
    :type 'integer
    :group 'eshell-ext)
  
! (defcustom eshell-explicit-command-char ?*
    "*If this char occurs before a command name, call it externally.
! That is, although vi may be an alias, *vi will always call the
! external version.  UNIX users may prefer this variable to be \."
    :type 'character
    :group 'eshell-ext)
  
--- 148,157 ----
    :type 'integer
    :group 'eshell-ext)
  
! (defcustom eshell-explicit-command-char ?\\
    "*If this char occurs before a command name, call it externally.
! That is, although vi may be an alias, \\vi will always call the
! external version."
    :type 'character
    :group 'eshell-ext)

Re: small Eshell docfix

[Miles Bader]

On Wed, Sep 01, 2004 at 10:23:07PM +0200, Emilio Lopes wrote:
> ! (defcustom eshell-explicit-command-char ?\\
>     "*If this char occurs before a command name, call it externally.
> ! That is, although vi may be an alias, \\vi will always call the
> ! external version."

Might want to put something like "MS Windows users may want to change this to
another character to avoid conflicting with the path-separator syntax" (just
to rub it in ... :-)

-Miles
-- 
"Though they may have different meanings, the cries of 'Yeeeee-haw!' and
 'Allahu akbar!' are, in spirit, not actually all that different."

Re: small Eshell docfix

[Kim F. Storm]

Miles Bader <miles@gnu.org> writes:

> On Wed, Sep 01, 2004 at 10:23:07PM +0200, Emilio Lopes wrote:
> > ! (defcustom eshell-explicit-command-char ?\\
> >     "*If this char occurs before a command name, call it externally.
> > ! That is, although vi may be an alias, \\vi will always call the
> > ! external version."
> 
> Might want to put something like "MS Windows users may want to change this to
> another character to avoid conflicting with the path-separator syntax" (just
> to rub it in ... :-)

[actually, in emacs terms, that is directory separator syntax :-) ]

But why don't we just let the default depend on system-type

        (if (memq system-type '(windows-nt ms-dos)) ?* ?\\)

Lots of other default settings already depend on system-type,
so I don't see why this is any different, as long as we get
the proper default for GNU.

Re: small Eshell docfix

[Emilio Lopes]

Kim F Storm writes:

> Miles Bader <miles@gnu.org> writes:

>> On Wed, Sep 01, 2004 at 10:23:07PM +0200, Emilio Lopes wrote:

>> > ! (defcustom eshell-explicit-command-char ?\\
>> >     "*If this char occurs before a command name, call
>> >     it externally.
>> > ! That is, although vi may be an alias, \\vi will
>> > always call the ! external version."
>>
>> Might want to put something like "MS Windows users may
>> want to change this to another character to avoid
>> conflicting with the path-separator syntax" (just to rub
>> it in ... :-)

Eshell doesn't accept `\' as directory separator.  An besides
the command interpreter in MS Windows doesn't has such a
function (quoting to protect from alias expansion).  So, for
MS Windows users there is no conflict or a tradition to be
broken here.

> But why don't we just let the default depend on
> system-type

>         (if (memq system-type '(windows-nt ms-dos)) ?*
>         ?\\)

> Lots of other default settings already depend on
> system-type, so I don't see why this is any different, as
> long as we get the proper default for GNU.

Well, one of the main advantages of Eshell is that its
behavior doesn't depend on the OS you are using.  You just
have the same interface everywhere.  To show the contents of
a directory you just use `ls', also on MS Windows (not `dir'!).

Having such a gratuitous difference would be confusing IMHO.

Re: small Eshell docfix

[John Wiegley]

Emilio Lopes <eclig@gmx.net> writes:

> Eshell doesn't accept `\' as directory separator.  An besides the
> command interpreter in MS Windows doesn't has such a function
> (quoting to protect from alias expansion).  So, for MS Windows users
> there is no conflict or a tradition to be broken here.

Actually, Eshell does accept \ as a directory separator on Windows
systems.

Now, thinking about all this has led me to the following proposal:

There are two different aspects to what "*" at the beginning of a
command word does: 1) It causes aliases to be ignored, and 2) it
prefers the external version of a program to any Lisp-implemented
version.

In the case of 1, I agree the default should be \.  In the case of 2,
I believe the current use of "*" ought to remain.  Thus I propose
creating a new variable, and a bit of new code, so that the meaning of
initial * is kept, while initial \ is allowed for ignoring aliases.
(The use of initial \ can be made safe for Windows users by parsing it
only if no other backslashes occur in the command word, and no such
executable exists in the root directory).

If the meaning of 1 and 2 is to stay merged, I recommend keeping the
current default of *, because it has a deeper meaning than what \ does
for shell users (where it means only that aliases be ignored).

John

Re: small Eshell docfix

[Eli Zaretskii]

> From: John Wiegley <johnw@gnu.org>
> Date: Sun, 05 Sep 2004 23:47:03 -0700
> Cc: rms@gnu.org, eclig@gmx.net
> 
> If the meaning of 1 and 2 is to stay merged, I recommend keeping the
> current default of *, because it has a deeper meaning than what \ does
> for shell users (where it means only that aliases be ignored).

FWIW, if you want to have OS-specific defaults for a character that
causes aliases to be ignored, I found that either a quote `"' or a
comma `,' will do that in CMD.EXE.

Re: small Eshell docfix

[Alex Schroeder]

"Eli Zaretskii" <eliz@gnu.org> writes:

> if you want to have OS-specific defaults for a character...

I lost track a bit.  Who's for OS-specific defaults?

I believe that OS-specific defaults are doing eshell users a
disservice.  I use both a GNU/Linux system at home and Windows at
work.  I use eshell on both machines regularly.  When I use eshell,
I'm not interested in OS-specific defaults.  I'm interested in the
uniform interface Emacs provides on both platforms.

Alex.
-- 
.O.  http://www.emacswiki.org/alex/
..O  Schroeder's fifth law:
OOO  Never accept more work than you can handle in one night of hacking.

Re: small Eshell docfix

[Eli Zaretskii]

> From: Alex Schroeder <alex@emacswiki.org>
> Date: Tue, 07 Sep 2004 18:29:50 +0200
> 
> "Eli Zaretskii" <eliz@gnu.org> writes:
> 
> > if you want to have OS-specific defaults for a character...
> 
> I lost track a bit.  Who's for OS-specific defaults?

I don't know (and don't care much ;-).

I understood that Richard decided that using `\' was the best solution
to whatever problem started this thread.  To that someone said that on
Windows, users cannot be expected to use `\'.  Someone else then said
that there are no aliases on Windows and no character to disable alias
substitution.  My message was the response to that last claim: there
_are_ aliases on Windows (through `doskey'), and the characters I
suggested disable alias expansion in that case.

I wrote that to give people who are discussing the matter information,
not to express my views on whether it is a good idea to use an
OS-specific character in this specific case.

> I believe that OS-specific defaults are doing eshell users a
> disservice.

I agree that, in general, OS-specific defaults should be avoided, but
Emacs does have them here and there.

Re: small Eshell docfix

[Alex Schroeder]

"Eli Zaretskii" <eliz@gnu.org> writes:

> I agree that, in general, OS-specific defaults should be avoided, but
> Emacs does have them here and there.

Well, to whomever this may concern, then, since you did not express
your views...  :)

Sometimes this makes sense, such as when we paste filenames containing
backslashes in response to find-file, or when we determine the name of
a helper program to use for some task.  The key is that these things
call programs of the underlying system, or work on strings generated
with the underlying system.  But that's not how eshell works.  We
would have to assume that eshell was used to run batch files or
something similar, and that those batch files used aliases or commands
starting with a '*' or something equally weird.

Sorry for wasting time, here, if all of this was obvious, or if the
decision has been made already.

Alex.
-- 
.O.  http://www.emacswiki.org/alex/
..O  Schroeder's fifth law:
OOO  Never accept more work than you can handle in one night of hacking.

Re: small Eshell docfix

[Richard Stallman]

The main question is simply this: which choice for the variable is
best for eshell users on the GNU system?  Is it * or \?

Re: small Eshell docfix

[John Wiegley]

Alex Schroeder <alex@emacswiki.org> writes:

> I believe that OS-specific defaults are doing eshell users a
> disservice.  I use both a GNU/Linux system at home and Windows at
> work.  I use eshell on both machines regularly.  When I use eshell,
> I'm not interested in OS-specific defaults.  I'm interested in the
> uniform interface Emacs provides on both platforms.

My feeling exactly.  If there is to be a default, I would like to see
it functional on all platforms.

Which begs the question: What is wrong with the current default?  All
this discussion seems to have sprung from an unfortunate reference to
UNIX in the docstring.  How about we just fix the docstring, and leave
the code alone?

John

Re: small Eshell docfix

[Miles Bader]

On Tue, Sep 07, 2004 at 04:28:36PM -0700, John Wiegley wrote:
> Which begs the question: What is wrong with the current default?  All
> this discussion seems to have sprung from an unfortunate reference to
> UNIX in the docstring.  How about we just fix the docstring, and leave
> the code alone?

Huh?  It's different that what other unix shells use -- people are used to
`\' suppressing aliases, so it's only natural that they'll try it in eshell.

As to your point about aliases vs. eshell builtins, I have no idea.  I'm not
sure whether people really care about the distinction or not (personally I
only use \ in specific well-known circumstances; given my restricted usage,
I'd probably be happy enough if \ only suppressed aliases).

-Miles
-- 
(\(\
(^.^)
(")")
*This is the cute bunny virus, please copy this into your sig so it can spread.

Re: small Eshell docfix

[John Wiegley]

Miles Bader <miles@gnu.org> writes:

> Huh?  It's different that what other unix shells use -- people are
> used to `\' suppressing aliases, so it's only natural that they'll
> try it in eshell.

"people" is the assumption in question.

Eshell has been using '*' as the alias escape character for five years
now.  Has it been a problem?

John

Re: small Eshell docfix

[Richard Stallman]

    But why don't we just let the default depend on system-type

	    (if (memq system-type '(windows-nt ms-dos)) ?* ?\\)

    Lots of other default settings already depend on system-type,
    so I don't see why this is any different, as long as we get
    the proper default for GNU.

It is fine to do this, if the best default is different for different
systems.

Re: small Eshell docfix

[Emilio Lopes]

Richard Stallman writes:

>     But why don't we just let the default depend on
>     system-type
> 	    (if (memq system-type '(windows-nt ms-dos)) ?*
> 	    ?\\)

>     Lots of other default settings already depend on
>     system-type, so I don't see why this is any different,
>     as long as we get the proper default for GNU.

> It is fine to do this, if the best default is different
> for different systems.

For users of DOS-like systems there is no "best" default,
since DOS command interpreters don't offer such functionality.

Indeed one of the advantages of Eshell is the uniform
interface across different systems, only possible because
it's written in Elisp.  Make it gratuitously behave
differently without any advantage from this change is
counter-productive for the users.

Re: small Eshell docfix

[Eli Zaretskii]

> From: Emilio Lopes <eclig@gmx.net>
> Date: Sat, 04 Sep 2004 22:06:32 +0200
> Cc: miles@gnu.org, emacs-devel@gnu.org, "Kim F. Storm" <storm@cua.dk>
> 
> For users of DOS-like systems there is no "best" default,
> since DOS command interpreters don't offer such functionality.

That's not true, AFAIK: Windows does offer the alias functionality
through `doskey'.

Could someone please look in the `doskey' documentation and see how
can one cause it to ignore an alias?