Front page | perl.perl5.porters |
Postings from March 2003
On 5 Mar 2003 06:02:33 -0000, perlbug-followup@perl.org wrote:
>On Mon, Feb 10, 2003 at 08:44:08PM -0000, Yitzchak Scott-Thoennes wrote:
>> BEGIN { local $@; die 'foo' } seems to bail out of the BEGIN, but doesn't
>> actually die, as if the BEGIN was an eval.
>
>The code in Perl_call_list checks for the length of ERRGV ( aka $@ ),
>and if this is zero, thinks all is OK and continues running the
>script.
>(see perl.c:3958).
I will take a look when I get a chance.
>There is no other way to catch an exception than checking if ERRGV isn't
>clear. Localizing it makes, of course, any exception un-catchable.
Not sure what you mean by catch/un-catchable here. Maybe it will be
clear when I look at the code.
>Are there any documentation that suggests that
> BEGIN { local $@; die ' ... ' }
>should die silently or something like that ?
Not that I can find.
>I ask it because many core modules seems to assume it and it actually
>_never_ worked like that. That assumption seems to cause lots of
>problems.
AFAICT from the documentation, the local should have
*no* effect on the die, and the die should end the program (unless the
BEGIN block is in a string eval, of course.)
Thread Previous
|
Thread Next