Tcl Assembly Language

[Andreas Kupries]

Karl Lehenbauer <ka...@NeoSoft.com> writes:

> Has anyone thought about creating a way for people to create new Tcl
> commands by directly coding bytecodes?

Yes, but not the time to do so.

--
Sincerely,
Andreas Kupries <a.ku...@westend.com>
<http://www.purl.org/NET/akupries/>
<http://www.tu-harburg.de/skf/tcltk/main.html> Coming ?
-------------------------------------------------------------------------------

[Paul Duffin]

Karl Lehenbauer wrote:
>
> Has anyone thought about creating a way for people to create new Tcl
> commands by directly coding bytecodes?
>

> It would be sort of a Tcl assembly language (Tal?). Besides being
> kinky, there could be some cool uses.
>
> DEC-10 LISP hackers might remember LAP, the LISP assembly program, which
> allowed you to write DEC-10 assembly code directly from LISP, with LISP
> syntax. Most Forths also had native reverse polish assemblers, too.
>
> Tal code, targeted to the Tcl virtual machine, would of course be
> hardware independent.
>

I would much rather have a mechanism for me to be able to byte compile my own
commands. This needs the ability to register a byte code as mine, with
associated callbacks etc and be called when my byte code is executed
with full access to the execution framework.

[Andreas Kupries]

Paul Duffin <pdu...@hursley.ibm.com> writes:

> Karl Lehenbauer wrote:

>> Has anyone thought about creating a way for people to create new
>> Tcl commands by directly coding bytecodes?

> I would much rather have a mechanism for me to be able to byte
> compile my own commands. This needs the ability to register a byte
> code as mine, with associated callbacks etc and be called when my
> byte code is executed with full access to the execution framework.

Ok, I added
http://www.purl.org/thecliff/tcl/wiki/820.html
(Bytecodes, Parsing and Execution)

to our wiki to collect these ideas.

[Jeffrey Hobbs]

Paul Duffin wrote:
>
> Karl Lehenbauer wrote:
> >
> > Has anyone thought about creating a way for people to create new Tcl
> > commands by directly coding bytecodes?
> >

> > It would be sort of a Tcl assembly language (Tal?). Besides being
> > kinky, there could be some cool uses.
> >
> > DEC-10 LISP hackers might remember LAP, the LISP assembly program, which
> > allowed you to write DEC-10 assembly code directly from LISP, with LISP
> > syntax. Most Forths also had native reverse polish assemblers, too.
> >
> > Tal code, targeted to the Tcl virtual machine, would of course be
> > hardware independent.
> >
>

> I would much rather have a mechanism for me to be able to byte compile my own
> commands. This needs the ability to register a byte code as mine, with
> associated callbacks etc and be called when my byte code is executed
> with full access to the execution framework.

I think that would be more interesting in the core, to extend the
current capabilities to allow such. At the moment it's all quite
static, but it may be possible to create dynamically extensible
byte codes and byte compiled commands.

--
Jeffrey Hobbs The Tcl Guy
hobbs at ajubasolutions.com Ajuba Solutions (née Scriptics)