At 09:29 AM 8/16/00 -0500, Jonathan Scott Duff wrote:
>Let me draw another picture (with "try" and "catch" this time):
>
> try {
> # code that may cause exceptions
> } catch { # catches ALL exceptions
> switch ($@->^_) {
> case canFoo { ... ; undef $@ }
> case isa('IOError') { ... ; undef $@ }
> case isa('Oops') { ... ; undef $@ }
> }
> } # implicit throw here if defined $@
>
>To signal that the exception was handled, the programmer undefs $@.
>That way if the user wishes to do something when an exception occurs,
>but not handle it, they just don't undef $@.
Well, this is the most credible proposal for using switch that I've yet
seen; but it forces the user to type the same piece of arcana into
practically every catch block. I want the commonest case to be the
simplest and most readable, and 99% of every exception handling thing I've
written has gone something like (adjusting to new syntax):
try {
blah; # usually tens of lines of code
} catch Exception::This {
blech; # handle Exception::This
} catch Exception::That {
plugh; # handle Exception::That
} catch {
xyzzy; # handle other exceptions (omitted 1/4 the time)
} finally {
foo; # Even this block only appears 1/2 the time or less
}
>BTW, I hope you guys don't find me frustrating,
Heck no, this is how it's supposed to work.
> I'm just trying to
>decide what exception handling should look like in Perl to me.
--
Peter Scott
Pacific Systems Design Technologies
