* Brian moseley ([EMAIL PROTECTED]) wrote:
> On Thu, 27 Dec 2001, Brian moseley wrote:
>
> > well I'm a victim of jet lag tonight, so here's a proposal
> > :) find it at
> > http://www.maz.org/perl/Serializer-0.01.tar.gz.
>
This is great! And for my tuppence[1] i think this is exactly what P5EE
needs - i.e. standards defined in Perl OO code along with working
interfaces to existing modules (after all those evil nasty Python
types are green with envy of CPAN[2] and we cannot lose that
advantage).
However I'm curious about a couple of things,
Where does Serialisation end and Persistent Objects begin? Sure
save/load[3] are equivalent to thaw/freeze but what about
lock/unlock[4], how will they integrate with the Serialisation
engines? Maybe a solution is to simply say that the PO engine
implements its own locking[5], this may seem a get out, however it may
just mean we lose a bit of optimisation and end up with a better
object model (although i may be wrong).
The second is how we can best represent standards for P5EE classes
inside Perl. This may be something that Damian might like to think
about or it may be something we can solve ourselves, however it seems
that if we end up creationg lots of modules with
sub foo {
die "Err, your implementation of Foo is not complete";
}
We will all go slightly mad.
Anyway thats my rather drunken comments, i'm sure i will be more
useful in the new year,
Greg
[1] http://bibliofile.mc.duke.edu/gww/EnglishMoney/Pre.html
[2] They may not be evil and they may not be nasty, but its always
good to have an enemy (Please refer to the classic 2000AD Nemesis
the Warlock strip where Torquemada decides to pick on "The
Frecks")
[3] Or whatever Chris' excellent definition called them.
[4] For whatever locking strategy you want.
[5] Via, the "container"/"oyster pot"/"whatever" object.
--
Greg McCarroll http://217.34.97.146/~gem/
