[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] Exception handling

Vincent Manis scripsit:

> No, I don't think so. I would like WG1 to consider (raise 4) illegal,
> as well as (raise x) for any value of x not produced by some variation
> of MAKE-ERROR-OBJECT. Then WG2 can decide to allow additional
> kinds of objects to be raised, without breaking compatibility with
> WG1. In my naïvete, I felt that making the action for (raise 4) be
> implementation-dependent would do this, but I'm happy with any other
> strategy that leads to the same result.

In my view that is too restrictive.  Allowing any object means that
WG1 users can roll their own condition types using DEFINE-RECORD-TYPE,
exporting the appropriate predicate and accessors from their modules,
without any bureaucratic restrictions.  Alternatively, they can use
association lists or the like.  This would not be possible if only
error objects were allowed: everything would have to be fitted into the
straitjacket of message + irritants.

La mayyitan ma qadirun yatabaqqa sarmadi                            John Cowan
Fa idha yaji' al-shudhdhadh fa-l-maut qad yantahi.              cowan@x
                --Abdullah al-Hazred, Al-`Azif      http://www.ccil.org/~cowan

Scheme-reports mailing list