[Larceny-users] Undefined binding not detected, uses it somehow

[Derick Eddington]

Larceny v0.97a4 (alpha test) (Mar 17 2009 11:29:20, precise:Linux:unified)
larceny.heap, built on Tue Mar 17 11:33:15 PDT 2009
ERR5RS mode (no libraries have been imported)

(import (rnrs base))
(define (f x)

(unless (number? x)
(assertion-violation 'f "not a number" x))
(- x))

(f 1)

Error: unhandled condition:
Compound condition has these components:
#<record &assertion>
#<record &who>
who : f
#<record &message>
message : "not a number"
#<record &irritants>
irritants : (1)

Entering debugger; type "?" for help.
debug> q

(number? 1)

#t

(unless (number? 1) 'blah)

Error: Undefined global variable "
unless".
Entering debugger; type "?" for help.
debug>

[Derick Eddington]

I realized this probably isn't a bug. It's evaluating the call to
assertion-violation before the reference to unless. I usually realize
sooner when this is happening. Sorry. I've been getting used to
libraries and programs which detect unbound identifiers sooner.

P.S. I can't find the part(s) of R6RS which talk about procedure call
sub-expressions evaluation order (i.e., that it's unspecified); where is
it? Thanks.
--
: Derick
----------------------------------------------------------------

Post by Derick Eddington
Larceny v0.97a4 (alpha test) (Mar 17 2009 11:29:20, precise:Linux:unified)
larceny.heap, built on Tue Mar 17 11:33:15 PDT 2009
ERR5RS mode (no libraries have been imported)

(import (rnrs base))
(define (f x)

(unless (number? x)
(assertion-violation 'f "not a number" x))
(- x))

(f 1)

#<record &assertion>
#<record &who>
who : f
#<record &message>
message : "not a number"
#<record &irritants>
irritants : (1)
Entering debugger; type "?" for help.
debug> q

(number? 1)

#t

(unless (number? 1) 'blah)

Error: Undefined global variable "
unless".
Entering debugger; type "?" for help.
debug>

[Jon Wells]

Page 28, second paragraph in the section titled "Procedure calls"...

When a procedure call is evaluated, the operator and
operand expressions are evaluated (in an unspecified or-
der) and the resulting procedure is passed the resulting
arguments.

jon.

Post by Derick Eddington
I realized this probably isn't a bug. It's evaluating the call to
assertion-violation before the reference to unless. I usually realize
sooner when this is happening. Sorry. I've been getting used to
libraries and programs which detect unbound identifiers sooner.
P.S. I can't find the part(s) of R6RS which talk about procedure call
sub-expressions evaluation order (i.e., that it's unspecified); where is
it? Thanks.

[Derick Eddington]

Thanks. I didn't guess it's in 9. Primitive syntax. I thought it would
be somewhere listed in the Contents with "procedure" in the name. Now I
see the section is referenced by the second index entry for "procedure
call", which I missed (I'll blame that on the HTML version). I note
R5RS has it in the "Procedure calls" section listed in the Contents.
--
: Derick
----------------------------------------------------------------

Post by Jon Wells
Page 28, second paragraph in the section titled "Procedure calls"...
When a procedure call is evaluated, the operator and
operand expressions are evaluated (in an unspecified or-
der) and the resulting procedure is passed the resulting
arguments.
jon.

Post by Derick Eddington
I realized this probably isn't a bug. It's evaluating the call to
assertion-violation before the reference to unless. I usually realize
sooner when this is happening. Sorry. I've been getting used to
libraries and programs which detect unbound identifiers sooner.
P.S. I can't find the part(s) of R6RS which talk about procedure call
sub-expressions evaluation order (i.e., that it's unspecified); where is
it? Thanks.