[m-rev.] for review: fix I/O retry bug

[Zoltan Somogyi]

On 11-Aug-2005, Ian MacLarty <maclarty at cs.mu.OZ.AU> wrote:
> Zoltan, I'm still not entirely sure why MR_trace_find_input_arg sometimes
> finds io.state values and sometimes doesn't.  Could you clarify that point if
> you know the reason?

Yes, I do.

If the compiler knows that an argument is of type io.state, it does not keep
its value around for events after the call, since the value is a dummy.
(In fact, values statically known to have type io.state aren't even passed
as arguments.) At later events therefore the RTTI won't include a reference
to the variable, and thus MR_trace_find_input_arg won't find it.

If the compiler knows the type of an argument only as a type variable T,
then the value probably isn't a dummy, and the compiler does make sure
that its value is available at later events (unless the argument has mode
di, in which case only events before its destruction will have access to it).
If the caller binds T to io.state, MR_trace_find_input_arg thus *will* find
the argument.

Zoltan.
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------