Front page | perl.perl5.porters |
Postings from July 2001
Graham Barr <gbarr@pobox.com> wrote:
|> > (it's just flipping a bit, so is very low impact).
|>
|> I am OK with adding, thats not my issue. My issue is with reading the hash.
|>
|> ie
|>
|> $var = $hash{element};
|>
|> should just continue to return undef, not croak. IMO
The feeling that motivated my wanting all this in the first place probably
was more concerned with prohibiting access to non-existant keys than with
creating new keys, since overall I'd guess that there are more accesses
than assignments. I'd certainly want to protect against typing
if ($hash{quite}) {
instead of
if ($hash{quiet}) {
just as I'd want to protect against
$hash{quite} = 1;
There's nothing stopping you from accessing any element you'd like (so
Perl's not forcing anything), but if you *do* want access limitiations to
what's currently there, I'd feel symmetrical restrictions make the most
sense.
And, since you can temporarily unclamp, it doesn't remove the functionality
you wish (it just makes it ugly). But to go the other way actually lowers
the overall functionality.
Here's an idea. It should be trivial to have two clamping bits:
clamp %hash, CLAMP_CREATE;
clamp %hash, CLAMP_ACCESS;
clamp %hash, CLAMP_ACCESS|CLAMP_CREAT;
clamp %hash, CLAMP_ALL; # same as CLAMP_ACCESS|CLAMP_CREAT
clamp %hash, CLAMP_NONE;
Jeffrey
Thread Previous
|
Thread Next