Re: Distributed Objects between cats
Re: Distributed Objects between cats
- Subject: Re: Distributed Objects between cats
- From: Wade Tregaskis <email@hidden>
- Date: Thu, 11 Aug 2005 21:09:10 +1000
Can I ask where you drew this conclusion from? I asked some fairly
knowledgeable folks in #macdev today and no-one had heard this was
this case. If it is it's pretty major news.....can you let us know
how you know this? Has Apple announced it?
It's half conjecture, half evidence. DO hasn't been updated in
forever. The design is a mess, which makes it very painful for 3rd
parties to extend it (I once did this, so I'm speaking from harsh
experience).
Also, back when I was playing with it, I was desperately seeking help
from anyone with knowledge of the DO internals. At some point, from
memory, someone (possibly from Apple) emailed me and said that
basically there was no-one at Apple anymore who knew much about DO;
they'd left or gone on to other projects.
Well actually, that's perhaps not entirely true - Quinn got back to
me at one point also, and said that Chris Kane and Doug Davidson are
worth asking about DO. They're both still at Apple afaik, and both
on this mailing list. I'm sure they'll pipe up now they've been
singled out. :)
In any case, the answers-to-questions ratio on the topic of DO is
generally very very low, which is perhaps the best indicator. I try
to answer any DO question that pops up, because of my affinity to it
and because I really love the concept, but the nature of most
problems are such that there's no simple solution.
Furthermore, these days compatibility with other platforms, languages
and environments is paramount, so things like DO have limited real
world applications. SOAP and XML-RPC and whatever else are all the
rage now. As much as I hate it, I'm looking now at moving one of my
current projects off DO, partly for these reasons (but primarily
because of security concerns*).
* = Which, btw, isn't to say there's any known issues with DO, but
it's too much of an unknown. It is a very advanced system, with a
lot of variability, and while I've the utmost respect for those that
wrote it, the design does have flaws**, which worry me.
** = And while I've mentioned this twice now, I can't really recall
too many specific examples off hand. One I do know is that despite
NSPort's being by nature high latency (i.e. network connections),
some parts of the DO ecosystem expect them to behave synchronously...
which you can overcome by prodding the runloop while effectively
blocking, but then parts of DO might accidentally get triggered off
other events, and at least one key part of the DO system is not re-
entrant like that... ad infinum. It doesn't sound all *that* bad as
a distant memory, but I remember wasting *months* trying to figure
out all the idiosyncrasies.
Wade Tregaskis (AIM/iChat, Yahoo, Gizmo & Skype: wadetregaskis, ICQ:
40056898, MSN: email@hidden, AV iChat & email:
email@hidden, Jabber: email@hidden)
-- Sed quis custodiet ipsos custodes?
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden