problem with srfi-1 when loading from the command line

problem with srfi-1 when loading from the command line

[Rick Taube]

Im having a problem getting the srfi-1 definition of list-index (which 
is defined in both guile and srfi-1) to stay in effect when I load my 
system from the command line using the -l option.  im not sure what is 
going wrong.

I load my program by loading a file called cm.scm, which in turn loads 
all my system's sources. The first file that it loads is called 
guile.scm which has the following use-modules in it:

---
(use-modules (srfi srfi-1))
(use-modules (oop goops))
(use-modules (ice-9 rdelim))
(use-modules (ice-9 pretty-print)) ; remove  at some point...
---

Now, if I start up Guile first and THEN  load cm.scm from inside it 
then everything works, as shown in this trace:
THIS WORKS:

galen:/Lisp/cm hkt$ guile
guile> (version)
"1.6.7"
guile> list-index
#<procedure list-index (l k)>                           ; <- the old 
list-index
guile> (load "/Lisp/cm/src/cm.scm")
; loading guile.scm
; loading loop.scm
; loading level1.scm
; loading utils.scm
; loading mop.scm
; loading objects.scm
; loading io.scm
; loading scheduler.scm
; loading sco.scm
; loading clm.scm
; loading clm2.scm
; loading midi1.scm
; loading midi2.scm
; loading midi3.scm
; loading data.scm
; loading scales.scm
; loading spectral.scm
; loading patterns.scm
guile> list-index
#<procedure list-index (pred clist1 . rest)>      ; <- the new one
guile> (quit)


NOW, if i load my program from the command line using -l option then it 
doesnt work, and adding --use-srfi=1  does not help (ive tried with and 
without the --use-srfi option)

THIS DOES NOT WORK:

galen:/Lisp/cm hkt$ guile --use-srfi="1" -l "/Lisp/cm/src/cm.scm"
; loading guile.scm
; loading loop.scm
; loading level1.scm
; loading utils.scm
; loading mop.scm
; loading objects.scm
; loading io.scm
; loading scheduler.scm
; loading sco.scm
; loading clm.scm
; loading clm2.scm
; loading midi1.scm
; loading midi2.scm
; loading midi3.scm
; loading data.scm
; loading scales.scm
; loading spectral.scm
; loading patterns.scm
guile> list-index
#<procedure list-index (l k)>      ; <- the old version is still set
guile> (quit)


As you can see, the list-index function was not redefined by srfi-1, or 
maybe it was and something that i dont understand is happeing. Any 
ideas what is going on? All the other definitions that i am using in 
srfi-1 are in effect. This  happens in guile 1.6.7 and 1.7.1

thanks
--rick



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

Re: problem with srfi-1 when loading from the command line

[Steve Juranich]

On 4/16/05, Rick Taube <taube@uiuc.edu> wrote:
> Im having a problem getting the srfi-1 definition of list-index (which
> is defined in both guile and srfi-1) to stay in effect when I load my
> system from the command line using the -l option.  im not sure what is
> going wrong.

This is kind of a knee-jerk reaction (I really should be in bed by
now), but perhaps guile is loading the SRFI module before it brings in
all of those "base" definitions.  I guess that doesn't make a lot of
sense, and I'm not sure how you'd prove that was happening.  Like I
said, I'm really tired. :-)

Good luck.
-- 
Steve Juranich
Tucson, AZ
USA


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

Re: problem with srfi-1 when loading from the command line

[Mike Gran]

Rick-

Tried it out.  I agree that it doesn't work, just as you described.

I was able to verify that the problem isn't in the order that things
are being defined.  The normal "list-index" gets defined first and then
the srfi-1 "list-index" gets defined later.  

Once the guile prompt comes up, "list-index", "map-in-order",
"for-each" and a bunch of other srfi-1 overwritten functions get
un-overwritten for some reason.  Don't know why.  Something about then
environment of the script being different from the environment of the
guile prompt?

Here's a workaround...  Call the following file with "guile -s tmp.scm"
instead of "guile -l tmp.scm"

---
;; tmp.scm

(use-modules 
        (srfi srfi-1)
        (oop goops)
        (ice-9 rdelim)
        (ice-9 pretty-print))

(scm-style-repl)
---

The "scm-style-repl" kicks off a new guile prompt in which the commands
at the prompt have the same environment as the script.  list-index
should then work as expected.

Hope this helps.

--
Mike Gran

--
> 
> As you can see, the list-index function was not redefined by srfi-1,
> or 
> maybe it was and something that i dont understand is happeing. Any 
> ideas what is going on? All the other definitions that i am using in 
> srfi-1 are in effect. This  happens in guile 1.6.7 and 1.7.1
> 
> thanks
> --rick
> 
> 
> 




		
__________________________________ 
Do you Yahoo!? 
Plan great trips with Yahoo! Travel: Now over 17,000 guides!
http://travel.yahoo.com/p-travelguide


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

Re: problem with srfi-1 when loading from the command line

[Marius Vollmer]

Rick Taube <taube@uiuc.edu> writes:

> Im having a problem getting the srfi-1 definition of list-index (which
> is defined in both guile and srfi-1) to stay in effect when I load my
> system from the command line using the -l option.  im not sure what is
> going wrong.

That is actually a 'feature' because of this code in boot-9.scm that
is executed just before the REPL starts:

    ;; so that builtin bindings will be checked first
    (module-use! guile-user-module (resolve-interface '(ice-9 r5rs)))
    (module-use! guile-user-module (resolve-interface '(guile)))

The builtin bindings come from the (guile), and using it last will
shadow all previous bindings.

I agree that this is not really wanted.  So I vote for removing the
last line of the code snippet above.  The (guile) module is used
anyway in every module.  Ok?

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


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

Re: problem with srfi-1 when loading from the command line

[Neil Jerram]

Marius Vollmer wrote:
> 
> I agree that this is not really wanted.  So I vote for removing the
> last line of the code snippet above.  The (guile) module is used
> anyway in every module.  Ok?
> 

OK with me.
	Neil


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