Hi Stef--
> Do you have a description of Naiad?
Not a proper one, no. There's just the Spoon mailing list archives[1]
and my "position statement" at Dan's request for the modules mailing
list[2]. Soon... I'm hoping we can still discuss what we want such a
system to do in the meantime, though.
> What is a Naiad module?
Currently, it's an object with:
- an ID
- added method IDs
- removed method IDs
- removed class IDs
- a name
- a description
- an author ID
- an initialization block
- prerequisite module IDs
- component module IDs
IDs are all Leach/Salz UUIDs. Two modules, each resident in different
object memories, can synchronize their respective systems via remote
messaging. The requesting module asks the providing module to
synchronize the two systems. The providing module:
- ensures that its author is defined in the requesting system
- has its prerequisite modules synchronize with the requesting system
- defines its added methods
- removes its removed methods
- removes its removed classes
- synchronizes its component modules with the requesting system
The requesting module, now equivalent to the providing module, runs the
initialization block. (Note that the receiving system can now act as a
provider to other systems.) The actual system changes made during the
synchronization depend on the initial state of the requesting system.
Many of the pitfalls of fileouts are avoided, since the author doesn't
have to second-guess the initial state of the requesting system. The
requesting and providing systems can work through their differences
almost entirely on their own. The information exchanged is not static,
as with a fileout or similar.
Requested methods can define their literals properly in the requesting
system, including defining new classes when necessary.
I've implemented a system for expressing Naiad commands as URLs. Such a
URL, when selected, issues a command to the Spoon system running on
localhost. The so-called "path" segment of a link is a text-encoded
action, which the local Spoon webserver evaluates.
There are several different commands: for quitting Spoon, making a
snapshot, installing modules, and so on. A module installation command
contains the hostname and port of the machine serving the module, and
the module's UUID. The receiving Spoon system uses this information to
create a remote-messaging connection between itself and the serving
system; the two systems then synchronize themselves via remote messages.
See [3] for an example of a Naiad link.
One simple way of enabling module discovery is simply to publish lists
of Naiad links on webpages, with module cataloging information, and let
Google do the indexing. I also intend to create a more integrated
discovery system with Spoon systems in a relay architecture, similar to IRC.
> I'm really interested in knowing that and what we could do with them +
> improving if necessary.
Now's a good time to discuss the design. What would you like it to do?
> Are other people welcome to use and build on top of spoon and Naiad?
Yes, see [4] (which has been in effect since the first public release,
of 14 February 2004).
> How do you see possible collaboration?
I'm all for it. :) Is there something more specific you want to know?
> > Ideally, I'd like the new modules to contain well-factored and
> > highly-readable expressions of the ideas of the old subsystems,
> > rather than just blind repackagings, but we'll see... it's tempting
> > to just imprint things to save time. :)
>
> You mean that you would like to rewrite some parts?
Yes, particularly the virtual machine implementation. I think it could
be much clearer in many places.
thanks!
-C
[1]
http://lists.squeakfoundation.org/pipermail/spoon[2]
http://minnow.cc.gatech.edu/squeak/5618[3]
http://netjam.org/spoon/modules[4]
http://netjam.org/spoon/releases/current/frame.php#license--
Craig Latta
improvisational musical informaticist
www.netjam.org
Smalltalkers do: [:it | All with: Class, (And love: it)]
Locked
