guile -s is it necessary?

guile -s is it necessary?

[tomas]

Hi,

as I'm making a fool of myself in these lists anyway, I might
well pose a question which nagged me for a while:

Is there any particular reason (besides of higher symmetry)
you have to invoke a script as ``guile -s <script name>''
instead of just ``guile <script name>''?

Note that the second form is not only convenient on the
command line but also simplifies your trustworthy hash-bang
line. As far as I know, unadorned arguments are not yet
used in guile, so this would be a backward-compatible
embellishment/enuglyment.

Opinions?

Regards

-- tomas


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Rob Browning]

tomas@fabula.de writes:

> Note that the second form is not only convenient on the
> command line but also simplifies your trustworthy hash-bang
> line. As far as I know, unadorned arguments are not yet
> used in guile, so this would be a backward-compatible
> embellishment/enuglyment.

What about things like:

  guile -s foo.scm arg-to-foo another-arg-to-foo

or

  guile -e main -s foo.scm arg-to-foo another-arg-to-foo

-- 
Rob Browning
rlb @defaultvalue.org, @linuxdevel.com, and @debian.org
Previously @cs.utexas.edu
GPG=1C58 8B2C FB5E 3F64 EA5C  64AE 78FE E5FE F0CB A0AD


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

Rob Browning <rlb@defaultvalue.org> wrote:
> What about things like:
>
>   guile -s foo.scm arg-to-foo another-arg-to-foo
>
> or
>
>   guile -e main -s foo.scm arg-to-foo another-arg-to-foo

guile foo.scm arg-to-foo another-arg-to-foo
guile -e main foo.scm arg-to-foo another-arg-to-foo

Such command lines are erroneous now, so it doesn't break anything to
make them mean the same as the -s versions (unless a different meaning
is already planned for the future).


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Neil Jerram]

>>>>> "tomas" == tomas  <tomas@fabula.de> writes:

    tomas> Is there any particular reason (besides of higher symmetry)
    tomas> you have to invoke a script as ``guile -s <script name>''
    tomas> instead of just ``guile <script name>''?

    tomas> Note that the second form is not only convenient on the
    tomas> command line but also simplifies your trustworthy hash-bang
    tomas> line. As far as I know, unadorned arguments are not yet
    tomas> used in guile, so this would be a backward-compatible
    tomas> embellishment/enuglyment.

    tomas> Opinions?

Sounds good to me (if there isn't some technical reason why we need
the -s).

        Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[tomas]

On Tue, Oct 15, 2002 at 04:40:01PM -0500, Rob Browning wrote:
> tomas@fabula.de writes:
> 
> > Note that the second form is not only convenient on the
> > command line but also simplifies your trustworthy hash-bang
> > line. As far as I know, unadorned arguments are not yet
> > used in guile, so this would be a backward-compatible
> > embellishment/enuglyment.
> 
> What about things like:
> 
>   guile -s foo.scm arg-to-foo another-arg-to-foo

Hmmm. First non-option argument would be the script name,
the rest is passed to it (the script) as arguments (that'd
mean that you must pass guile-specific switches *before*
the first non-option argument. This would again match the
two BigPee's behaviour, whether this be considered a Good
Thing or not ;)

> or
> 
>   guile -e main -s foo.scm arg-to-foo another-arg-to-foo

Both BigPee's know a switch like that for an expression to be
evaluated. Perl (-e) takes several of them and a script on top
of that, Python just one (-c) and terminates options processing
(matching the more cautious nature of Pythoners in general ;^)

I think guile's -e is special in that it is `out of order',
first loading the script and then calling the function passed
as argument; so the more exact correspondend to Perl's -e and
Python's -c would be guile's -c -- but that wouldn't change
the `syntax' of the command line.

Thanks
-- tomas


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Neil Jerram]

>>>>> "Paul" == Paul Jarc <prj@po.cwru.edu> writes:

    Paul> Rob Browning <rlb@defaultvalue.org> wrote:
    >> What about things like:
    >> 
    >> guile -s foo.scm arg-to-foo another-arg-to-foo
    >> 
    >> or
    >> 
    >> guile -e main -s foo.scm arg-to-foo another-arg-to-foo

    Paul> guile foo.scm arg-to-foo another-arg-to-foo
    Paul> guile -e main foo.scm arg-to-foo another-arg-to-foo

Indeed.  Guile's current behaviour is to stop parsing args itself when
it sees `-s sthg', so that the remaining args are left for the
script.  We could easily make it do the same when it sees `sthg' where
sthg is not a known option.

    Paul> Such command lines are erroneous now, [...]

Why do you say they are erroneous?  They look fine to me.

        Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

Neil Jerram <neil@ossau.uklinux.net> wrote:
>>>>>> "Paul" == Paul Jarc <prj@po.cwru.edu> writes:
>     Paul> guile foo.scm arg-to-foo another-arg-to-foo
>     Paul> guile -e main foo.scm arg-to-foo another-arg-to-foo
>     Paul> Such command lines are erroneous now, [...]
>
> Why do you say they are erroneous?  They look fine to me.

$ guile foo.scm arg-to-foo another-arg-to-foo
guile: Unrecognized switch `foo.scm'
[...]

I.e., guile treats this kind of command line as an error.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

tomas@fabula.de wrote:
> Is there any particular reason (besides of higher symmetry)
> you have to invoke a script as ``guile -s <script name>''
> instead of just ``guile <script name>''?

One might argue that someone invoking "guile foo" might have intended
"guile -- foo" instead of "guile -s foo", so "guile foo" is a bit
error-prone.  It's less likely that someone would accidentally supply
an explicit -s when they meant --.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Marius Vollmer]

prj@po.cwru.edu (Paul Jarc) writes:

> One might argue that someone invoking "guile foo" might have intended
> "guile -- foo" instead of "guile -s foo", so "guile foo" is a bit
> error-prone.  It's less likely that someone would accidentally supply
> an explicit -s when they meant --.

Of what use is "guile -- foo"?  It will start the repl with 

    guile> (command-line)
    ("guile" "foo")

Is that useful?

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Marius Vollmer]

tomas@fabula.de writes:

> Is there any particular reason (besides of higher symmetry)
> you have to invoke a script as ``guile -s <script name>''
> instead of just ``guile <script name>''?

I don't know of any.  Allowing "guile <script name>" seems very
natural and rhymes with Perl, Python, Sh, and probably others.

There don't seem to be real objections, so: do you have a patch? :-)

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

Marius Vollmer <mvo@zagadka.ping.de> wrote:
>     guile> (command-line)
>     ("guile" "foo")
>
> Is that useful?

Not very, I imagine.  I'm not opposed to making -s optional; I was
just speculating as to why it wasn't already.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Neil Jerram]

>>>>> "Marius" == Marius Vollmer <mvo@zagadka.ping.de> writes:

    Marius> Allowing "guile <script name>" seems very natural and
    Marius> rhymes with Perl, Python, Sh, and probably others.

Rhymes?  Only if <script name> is "curl" or "siphon" (and even then
only questionably).

Perhaps you meant chimes?

With apologies for the irrelevant nature of this post ... :-)

        Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Marius Vollmer]

Neil Jerram <neil@ossau.uklinux.net> writes:

> >>>>> "Marius" == Marius Vollmer <mvo@zagadka.ping.de> writes:
> 
>     Marius> Allowing "guile <script name>" seems very natural and
>     Marius> rhymes with Perl, Python, Sh, and probably others.
> 
> Rhymes?  Only if <script name> is "curl" or "siphon" (and even then
> only questionably).
> 
> Perhaps you meant chimes?

:-) Hehe, sometimes, too often, I just use words and phrases where I
don't really know what they mean and make a mental note to check them
before sending the mail.  I don't do the checking, of course.

Do you really say "syphon"?  Or is it "pithon"?  To my foreign ears,
siphon and python don't sound the same... :-) (Ahh, I just checked, it
really is "syphon".  We say "siffon".)

But I did _spell_ "rhyme" right!

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

tomas@fabula.de wrote:
> Is there any particular reason (besides of higher symmetry)
> you have to invoke a script as ``guile -s <script name>''
> instead of just ``guile <script name>''?

Ok, I thought of another possible problem.  Suppose a program wants to
invoke guile to process a script.  If -s is required, then the program
will use it.  If -s is optional, the programmer will be tempted to use
the plain "guile $script_name" form, which will be buggy if
script_name can start with "-".  Also consider code that parses a
guile command line - say, a wrapper around guile.  That code will also
have to be updated to accomodate making -s optional.

As it is, we can know what each argument is (i.e., a switch, a script
name, or a script argument) without looking at the contents of the
argument; we only need to look at the preceding arguments.  I think
that's valuable.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Marius Vollmer]

prj@po.cwru.edu (Paul Jarc) writes:

> Ok, I thought of another possible problem.  Suppose a program wants
> to invoke guile to process a script.  If -s is required, then the
> program will use it.  If -s is optional, the programmer will be
> tempted to use the plain "guile $script_name" form, which will be
> buggy if script_name can start with "-".

Yes, that's a good argument in favor of using "-s".  The documentation
of the option-less way of running a script should definitely talk
about this.

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Robert Uhl <ruhl@4dv.net>]

Marius Vollmer <mvo@zagadka.ping.de> writes:
> 
> Do you really say "syphon"?  Or is it "pithon"?  To my foreign ears,
> siphon and python don't sound the same... :-) (Ahh, I just checked,
> it really is "syphon".  We say "siffon".)

We say sigh-fin, as opposed to pie-thahn or pyth-ahn.

-- 
Robert Uhl <ruhl@4dv.net>
Those who beat their swords into ploughs will plough for those that don't.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Robert Uhl <ruhl@4dv.net>]

prj@po.cwru.edu (Paul Jarc) writes:
> 
> Ok, I thought of another possible problem.  Suppose a program wants
> to invoke guile to process a script.  If -s is required, then the
> program will use it.  If -s is optional, the programmer will be
> tempted to use the plain "guile $script_name" form, which will be
> buggy if script_name can start with "-".

Granted--but the same problem exists with everyone else.  No sense
being gratuitously incompatible.  And of course the wise programmer
would write "guile -- $script_name" (or whatever).

-- 
Robert Uhl <ruhl@4dv.net>
When any government, or any church for that matter, undertakes to say
to its subjects, `This you may not read, this you must not see, this
you are forbidden to know,' the end result is tyranny and oppression,
no matter how holy the motives.  Mighty little force is needed to
control a man whose mind has been hoodwinked; contrariwise, no amount
of force can control a free man, a man whose mind is free.  No, not the
rack, not fission bombs, not anything.  You can't conquer a free man;
the most you can do is kill him.               --Robert A.  Heinlein


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Joshua Judson Rosen]

[-- Attachment #1: Type: text/plain, Size: 617 bytes --]

On Sat, Oct 19, 2002 at 12:07:56AM +0200, Marius Vollmer wrote:
> tomas@fabula.de writes:
> 
> > Is there any particular reason (besides of higher symmetry)
> > you have to invoke a script as ``guile -s <script name>''
> > instead of just ``guile <script name>''?
> 
> I don't know of any.  Allowing "guile <script name>" seems very
> natural and rhymes with Perl, Python, Sh, and probably others.

And it'll let me use "#!/usr/bin/env guile" at the top of my scripts :)

-- 
"Great, I'm talking to a mathmetical symbol. I'm staying away from
 the week old pizza served across the street." --Kane, TCD

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: guile -s is it necessary?

[Rob Browning]

ruhl@4dv.net (Robert Uhl <ruhl@4dv.net>) writes:

> We say sigh-fin, as opposed to pie-thahn or pyth-ahn.

or sy-fun if you're from further south.

-- 
Rob Browning
rlb @defaultvalue.org, @linuxdevel.com, and @debian.org
Previously @cs.utexas.edu
GPG=1C58 8B2C FB5E 3F64 EA5C  64AE 78FE E5FE F0CB A0AD


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

ruhl@4dv.net (Robert Uhl <ruhl@4dv.net>) wrote:
> prj@po.cwru.edu (Paul Jarc) writes:
>> program will use it.  If -s is optional, the programmer will be
>> tempted to use the plain "guile $script_name" form, which will be
>> buggy if script_name can start with "-".
>
> Granted--but the same problem exists with everyone else.

Right, but I don't think adding to the problem is useful.

>  No sense being gratuitously incompatible.

It's not an incompatibility; it's a dissimilarity.  No help being
gratuitously similar.  The current behavior of requiring -s breaks
nothing; it's merely unexpected by people accustomed to sloppier
interfaces of other interpreters.

> And of course the wise programmer would write "guile --
> $script_name" (or whatever).

Only if they didn't want to run the script.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

Joshua Judson Rosen <rozzin@geekspace.com> wrote:
> And it'll let me use "#!/usr/bin/env guile" at the top of my scripts :)

You can already get the same effect with even more flexibility; see
the end of the "Executable Modules" node in the manual.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Robert Uhl <ruhl@4dv.net>]

Rob Browning <rlb@defaultvalue.org> writes:
> 
> > We say sigh-fin, as opposed to pie-thahn or pyth-ahn.
> 
> or sy-fun if you're from further south.

Well, the second syllable's vowel is a schwa--which means it can run
the gamut of sounds, just about.  I say it with something between i &
u.

-- 
Robert Uhl <ruhl@4dv.net>
Marmite is a black tarry yeast extract used to encourage New Zealand children
to grow large and strong.  The usual method is to feed it to them on toast,
resulting in them growing very quickly, so that they might become big and
strong enough to stop their parents from doing this.      --Rupert Boleyn


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

guile -s is it necessary?

[Sven Hartrumpf]

Hi.
If you want to call Scheme scripts more portably, you should probably use
SRFI-22 (but guile does not support this SRFI, yet?):
http://srfi.schemers.org/srfi-22/srfi-22.html
Greetings
Sven


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Paul Jarc]

Sven Hartrumpf <Sven.Hartrumpf@FernUni-Hagen.de> wrote:
> If you want to call Scheme scripts more portably, you should probably use
> SRFI-22 (but guile does not support this SRFI, yet?):
> http://srfi.schemers.org/srfi-22/srfi-22.html

I don't think it does; srfi-22 is incompatible with the existing
(and more useful, I think) interpretation of #! borrowed from scsh.
Well, almost.  Guile could decide to use the srfi-22 interpretation
when the first argument is a script name, and keep the current
behavior otherwise.  But that would be awfully ugly.  If srfi-22 is to
be supported, I'd suggest instead providing a separate guile-srfi-22
executable which accepts a script name as the first argument without
-s, and which interprets #! as a comment ending at the next newline.


paul


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Rob Browning]

prj@po.cwru.edu (Paul Jarc) writes:

> If srfi-22 is to be supported, I'd suggest instead providing a
> separate guile-srfi-22 executable

That's my current inclination as well.

-- 
Rob Browning
rlb @defaultvalue.org, @linuxdevel.com, and @debian.org
Previously @cs.utexas.edu
GPG=1C58 8B2C FB5E 3F64 EA5C  64AE 78FE E5FE F0CB A0AD


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Neil Jerram]

>>>>> "Rob" == Rob Browning <rlb@defaultvalue.org> writes:

    Rob> prj@po.cwru.edu (Paul Jarc) writes:
    >> If srfi-22 is to be supported, I'd suggest instead providing a
    >> separate guile-srfi-22 executable

    Rob> That's my current inclination as well.

This isn't the best thread to say this in, but ...

In general, I think we should _not_ jump to implement every SRFI that
gets through the process.  Unfortunately, IMO, the level of review in
that process is not always top quality, and it is too easy for
anything to get through if the author just pushes enough.

In particular, as regards SRFI-22, I think it is regrettable that

- the SRFI has no consideration for non-Unix-like systems

- Guile's existing usage of #! is ignored.

        Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: guile -s is it necessary?

[Rob Browning]

Neil Jerram <neil@ossau.uklinux.net> writes:

> This isn't the best thread to say this in, but ...
>
> In general, I think we should _not_ jump to implement every SRFI that
> gets through the process.  Unfortunately, IMO, the level of review in
> that process is not always top quality, and it is too easy for
> anything to get through if the author just pushes enough.

Makes sense, though if we can do it in a way that isn't hard and
doesn't break things on our side just to be compatible, then providing
compatibility in some form probably wouldn't bother me.  That's why in
this case I was leaning toward the separate name (if we did anything).

I do agree, though, that we shouldn't feel compelled to implement
every SRFI, and perhaps that applies for this one -- I'll have to go
look at it again -- all I recal ATM is that from the previous
discussion, a separate executable seemed preferable to trying to
integrate its requirements with our current command line and #!
practices.

-- 
Rob Browning
rlb @defaultvalue.org, @linuxdevel.com, and @debian.org
Previously @cs.utexas.edu
GPG=1C58 8B2C FB5E 3F64 EA5C  64AE 78FE E5FE F0CB A0AD


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user