On 11 Jul 2002, at 15:02, Jeff Zucker wrote:
> Pomin Wu wrote:
>
> > When accessing CSV with AnyData as a tie hash, I found
> > that although the C<while (my $row = each %$table)> syntax
> > is valuable, it C<last> as soon as I try to modify the
> > table inside the loop, for ex. C<delete $table->{key};>.
> > Or have I missed something?
>
>
> Not sure what you mean. This works:
>
> use AnyData;
> my $table = adTie( 'CSV', ["id,phrase\n1,foo\n2,bar\n3,baz"] );
> while (my $row=each %$table) {
> delete $table->{$row->{id}}
> if $row->{id}<3 and $row->{phrase} ne 'bar';
> }
>
> But if all you want to do is delete a set of rows, you don't need the
> while loop, you can just do something like this which accomplishes the
> same thing as the while loop above:
>
> delete $table->{{id=>'< 3',phrase=>'ne bar'}};
>
> If I've misunderstood your question, please send an example of the
> offending code.
>
> --
> Jeff
>
Although it may work for this particular case, I consider it critical
to edit hashes while iterating over them with the while/each
construct. Although deleting is "officially" allowed (see perlfaq4)
for conventional hashes since perl5, and changing the current value
is also possible, this can get you into trouble with tied hashes
(e.g., changing the values of a hash tied to MLDBM during iteration
with while/each can bomb you into an infinite loop). In general, I
think that while/each is for fast read-only access, and for/keys for
editing purposes.
Bodo
