RE: [ANN] LuaCSP framework: Communicating Sequential Processes in Lua

[Alexey Baskakov <alexey_baskakov@...>]

> Like I said, you can check out my fork on github (I'm not going to
> bother to submit a pull request unless you really want me to.) I've
> already got it compiling (manually with g++.) 
Thanks, I'll check it out later!

> One thing which might be
> useful is using CMake since it can generate MVS solutions as well as
> Makefiles (or Ninja builds, which is becoming my preferred builder now.)
> Otherwise a plain Makefile is really good enough.
Yeah, cmake is mentioned in Development Roadmap section :)


> I guess the way I'd /like/ to use it would be two fold (depending on
> what I'm writing.) One would be the embedded route which is currently
> supported. The other would be more in pure Lua (or using C modules)
> being run from the standard Lua interpreter. Some things I'd like to be
> able to do:
> * Use channels and composition to make some code concurrent.
> * Write fundamental operations in plain Lua.
Yeah, fundamental operations in Lua are in my todo list. And I need them quite urgently.
It'll be like this:

OpLuaSleep = CspOperation:new()

function OpLuaSleep:Work( deltaTime )
	self.seconds = self.seconds - deltaTime
	if self.seconds > 0 then
		return CspOperation.Yield
	else
		return CspOperation.Finish
	end
end

function OpLuaSleep:Terminate()
end

function LUA_SLEEP( seconds )
	local op = OpLuaSleep:new()
	op.seconds = seconds
	return op.DoInit()
end

Of course, you need some async API in plain Lua/C/C++ to wrap into fundamental operations...


> What I guess I'd like to be able to write is something like:
> http://pastebin.com/DQPZPPr9

In general, you should be able to pass any number of arguments of any type to CSP process (including non-CSP closures).
Since Swarm:go can't distinguish whether a closure is a process to start or an argument, it's more reasonable to use two anonymous
closures as in the doc.

Of course, there is a variant to organize API like this:
swm:go( p1, {arg1, arg2}, p2, {arg3, arg4} )
but I found this less convenient than extra anonymous closures (from my experience). 

In fact, Swarm:go isn't the key feature and a bad practice (maybe). It just slightly simplifies the code at some cases.
It's provided for fun, as a show of respect to Google Go language :)
However, all the mentioned details are applicable to PAR as well.

BTW, I want to make ALT to act as 'select' in Go. I.e. to be able to wait not only on channel inputs but outputs too.

Alexey.