Re: [Chicken-users] chicken-hosted stalin

[felix winkelmann]

On Jan 22, 2008 3:45 PM, John Cowan <address@hidden> wrote:
> felix winkelmann scripsit:
>
> > Even though the [chicken-stalin] compiler is even slower than the
> > original one, it compiles (the compiler itself) faster [...].
>
> This is indecipherable.  The purpose of a compiler is to compile; if
> chicken-stalin compiles the Stalin source code faster than native stalin
> does, in what sense is it also slower?  And if chicken-stalin *is* slower,
> what is the point of compiling Stalin with csc, since a program that
> is inherently as slow as Stalin should presumably be compiled with the
> compiler that produces the fastest executables, viz. native stalin?

chicken-stalin can be built on systems that are too weak to build
the regular stalin, it also runs probably on more platforms.
I think this is useful. Due to the reduced turnaround times and the much
richer runtime environment, one can actually debug and enhance
the system (much more so as the original one).

>
> Also, I'm not clear how much use Stalin is in the Chicken environment,
> given its severe restrictions on input language.  If Stalin could
> be front-ended with an R5RS macro expander, and if enough bootstrap
> procedures could be provided that it could understand a larger fraction of
> Chicken Scheme, then it would indeed be useful to have as an alternative
> compiler.

You have to see for yourself whether you find it useful. It is possible
to mix chicken and stalin code, so one could conceive trying to get the
best of both worlds (partially fast code and partically using a rich
and dynamic runtime environment).

Adding macros and more compiler language support could be done, though.
So this is a start.


cheers,
felix