[Twisted-Python] Hacking Reality

Chris Armstrong carmstro at twistedmatrix.com
Wed Oct 10 00:03:25 EDT 2001


On Wed, Oct 10, 2001 at 10:21:20AM +1000, Chris.Hadgis at mincom.com wrote:
> I would be interested in helping to develop reality further.
> 
> Where do I sign?
> 
> ChrisH, back after two and a half weeks away from computers

Well, your timing is interesting. I just got a phone call last night from
Glyph, and we talked a bit about how the new spatial simulation system is
_not_ to be developed inside twisted.reality, but as a separate package.
Twisted.reality still does need a lot of work, and glyph pointed out a few
things to me that need done:

1) Re-do the broadcasting system. It's kind of a mess now. I've taken a
_look_ at this, but I'm not really sure where to take it. IIRC you were the
guy who was interested in this before, so if you could do a little work in
this area it'd be great.

2) Get rid of the 'reality' reference inside things. Glyph says this would
apparently reduce memory footprint by a huge amount. I think I'll analyze
exactly how much work getting rid of this is going to mean. Obviously it's
not just a simple "ok, delete all code using thing.reality" -- it's a big
structural change (Realities should reference things, and not the other way
around)

3) Get rid of the global object dicts. This would probably be a lot of work
as a lot of code would get touched by this. As glyph has preached before,
"Twisted represents a huge directional graph" -- things can get references
to each other any number of ways, without a global thing store.

So, now that Reality and ImaginarySpatialSim are no longer together, I'd
like to know how the heck we're going to do this. Jason, any thoughts?
I still have some doubt that there is no common ground between a textual and
a spatial simulation, and I still dream of a game that both textual and
spatial representation of objects is possible (I suppose this would be
possible even without a tighter "integration" of reality and SpatialSim, ie,
class MyThingy(reality.Thing, spatial.Thing)). The ideas that Jason has told
me about in the past seem to promising for me to give up the idea of an
integrated "simulation" system for Twisted (even though Glyph professes to
me that Twisted _is_ the simulation system, but...). 

Ehh, that last paragraph was pretty much a braindump. But I am kinda tired
of being completely clueless about how all this stuff is gonna fit together.

--
Chris Armstrong  |  carmstro at twistedmatrix.com  |  cjast3 at robert-morris.edu
Web Page: http://purl.org/net/ChrisArmstrong





More information about the Twisted-Python mailing list