conditions
Fredrik Sandstrom
Suppose I have the following form:
(destructuring-bind (a b c) list
...)
What I want to do is handle the error of LIST not matching the given
lambda list. The problem is, how do I know what error to look for?
CMUCL signals a condition of type
COMMON-LISP::DEFMACRO-LL-ARG-COUNT-ERROR which is helpful enough, but
CLISP just signals a SIMPLE-ERROR. Furthermore, I do not want to trap
errors that come from the body of the destructuring-bind form, only
errors with the actual binding of values to variables. How do I do this,
short of writing my own routine that checks if the list is OK?
--
Fredrik Sandström
fre...@infa.abo.fi
Paul F. Dietz
> Suppose I have the following form:
> (destructuring-bind (a b c) list
> ...)
>
> What I want to do is handle the error of LIST not matching the given
> lambda list. The problem is, how do I know what error to look for?
> CMUCL signals a condition of type
> COMMON-LISP::DEFMACRO-LL-ARG-COUNT-ERROR which is helpful enough, but
> CLISP just signals a SIMPLE-ERROR. Furthermore, I do not want to trap
> errors that come from the body of the destructuring-bind form, only
> errors with the actual binding of values to variables. How do I do this,
> short of writing my own routine that checks if the list is OK?
The standard merely requires DESTRUCTURING-BIND to signal an ERROR
in this case; anything more detailed is implementation-specific.
You can avoid having the body's errors be caught by doing this:
(multiple-value-bind (a b c)
(handler-case (destructuring-bind (a b c) list
(values a b c))
;; error handling cases for destructuring-bind here
)
;; original body of destructuring-bind here
)
You can write a macro to generate this.
Paul
Frode Vatvedt Fjeld
> I'm having problems getting the hang of the condition system...
>
> Suppose I have the following form:
> (destructuring-bind (a b c) list
> ...)
>
> What I want to do is handle the error of LIST not matching the given
> lambda list. The problem is, how do I know what error to look for?
According to CLHS, all you can depend upon is a condition of type
error, which won't tell you much. I think this means that d-bind is
only intended to be used when you believe you know the structure of
the list being destructured, and not as a matching or unification
probe. Then you have to write your own parser, or use some library.
> Furthermore, I do not want to trap errors that come from the body of
> the destructuring-bind form, only errors with the actual binding of
> values to variables.
That's not so difficult, just rewrite
(destructuring-bind (<lambda-list>) <form> <body>)
to something like
(multiple-value-bind (<lambda-variables>)
(handler-case
(destructuring-bind (<lambda-list>) <form>
(values <lambda-variables>))
...)
<body>)
--
Frode Vatvedt Fjeld
Paul F. Dietz
> The standard merely requires DESTRUCTURING-BIND to signal an ERROR
> in this case; anything more detailed is implementation-specific.
More accurately, it only requires that the error be signaled in
safe code.
Paul
Will Fitzgerald
only "should" signal an error, which means your code can be guaranteed to
catch an error when safety optimization is set to 3 (i.e., "safe code").
I think this is what is needed, then:
(multiple-value-bind (<lambda-variables>)
(handler-case
(destructuring-bind (<lambda-list>) <form>
(declare (optimize (safety 3)))
(values <lambda-variables>))
...)
<body>)
"Frode Vatvedt Fjeld" <fro...@acm.org> wrote in message
news:2hadn8a...@vserver.cs.uit.no...