I'm spending my lunch hour re-reading and absorbing old info, trying to
get back in the Frontier development mindset. I had some thoughts so I
figured that I would push something out here and maybe follow up with a
blog post later. What follows is not intended to be a declaration but
instead some type of conversation starter.

Frontier is:

* an outliner
* an interactive UserTalk runtime
* a virtual machine with it's own rules
* a way to glue ideas together and automate results
* a graphical view of code

So I was thinking about all of these things and it made me wonder, does
the app that drives the GUI have to also drive the runtime? I'm
thinking of things like a weblog editing app (MarsEdit) that talks to a
blogging platform (like WordPress). Two things that work together to
produce results but are "loosely coupled". What about something as
simple as a GUI for the UNIX cron daemon? Wouldn't that be similar
(albeit simplistic?) to what the kernel does for us today? The daemon
runs in the background, waiting for events to be triggered and if you
have a GUI for the config files, you could write shell scripts that
would then be executed with results.

Now that I've established a baseline of thinking, here's what I'm
getting at:

Frontier could consist of many parts, loosely coupled, but working
together based on that person's needs:

1. A kernel daemon that would do string parsing, UserTalk complilation,
and data storage and manipulation
2. A GUI app that would be an outliner with an "offline/online"
mentality, much like Radio used to talk to a Manila ODB.

Enough for this email. I'll try to post something on my weblog as I
sort this out.

Steve