What's a dump stream actually going to look like? E.g.,
$a = [1, 2, 3];
print dump($a);
1: PMC PerlArray
2: 3 # element count
3: PMC PerlInt # [0]
4: 1 # value
5: PMC PerlInt # [1]
6: 2 # value
7: PMC PerlInt # [2]
8: 3 # value
Fine and good. But dump must handle object graphs, which can be
recursive or self-referential or whatnot:
$a = { b => undef };
$b = { a => $a };
$a->{b} = $b;
print dump($a);
1: PMC PerlHash
2: 1 # key-value count
3: PMC PerlString # keys[0]
4: "b" # value
5: PMC PerlHash # values[0]
6: 1 # key-value count
7: PMC PerlString # keys[0]
8: "a" # value
9: !!!!!!!
Panic! Dump needs to refer to line 4 again, or else recurse! How?
Most serializers use a seen hash not only to indicate that an object has
been seen, but also to remember a name or number for that object. If
dump had remembered &$a => line 4, it could finish off with something
like...
9: Again 4 # values[0]
A seen hash is also threadsafe, can be interrupted by DoD, and is safe
for recursion. (By "threadsafe," I mean that unchanged data structures
can be safely serialized from multiple threads without ill effect or
possibility of deadlock.)
--
Gordon Henriksen
IT Manager
ICLUBcentral Inc.
[EMAIL PROTECTED]
