Re: GNU Extension Language Plans

John Ousterhout (ouster@tcl.eng.sun.com)
2 Nov 1994 18:46:19 GMT

In article <Yigde=200aw3M_udcN@andrew.cmu.edu>, "Daniel C. Wang" <dw3u+@andrew.cmu.edu> writes:
|>
|> The fact that there isn't a bytecode compiler for TCL, should be a good
|> indication that it's a non-trival problem. Someone at was working on one for
|> his master thesis, I've forgoten exactly what happend to it, but my
|> impression was that the semantics of TCL made it hard to compile to bytecode
|> in an efficent way.
|>

The fact that something doesn't exist doesn't necessarily mean it's hard.
I used to think that a compiler for Tcl was hard, but Adam Sah's M.S.
project, referred to above, convinced me that it's not really that hard
at all. The only reason that Adam's work didn't find its way into Tcl
is that it wasn't organized right: it compiled statically rather than
on-the-fly, and this aspect makes his compiler clumsy to use in Tcl.
Doing a good compiler is getting higher and higher on my priority list,
but doing the PC and Mac ports are quite a bit higher still.