My final 5 cents.
> ...  Python ...
Python is a very nice language, but ... does one really need it's power ?
A quick look shows - it's source is 2.2MB gziped ( LyX 1.0 is 1.9MB
gziped ). Moreover, it's constantly beeing enhanced ...
One also needs to write a glue code between LyX's C/C++ code and Python.
> ... CINT C/C++ interpreter ...
The source code is less then 1MB gziped.
One could make ( automatically ) a "dictionary" of existing LyX classes,
objects, functions, variables, ... making them visible to the interpreter.
Any script would be simply a piece of C/C++ code ( one does not need to
know all "extras" of C++ to use it ).
> ... S-Lang ...
The source code is about 500kB gziped.
Simpler C-like syntax. Already used in an editor environment ( I would
like to remind you that the Ivy was also planned to be an extention
language for the "joe" editor, as far as I remember ).
> ... Guile ... SIOD ... elisp ...
All of them are lisp - tons of parentheses, magic behaviour, ... - as long
as you know lisp it's fine, but a novice user will have problems - let's
ask how many people, who use emacs, are able to write elisp scripts ? I
doubt there will be more in LyX's case ( note that even creators of emacs
by default disable "eval-expression ( ESC-ESC )" in order to "protect" the
avarage user from lisp ). So, I doubt the lisp-like scripts are good idea.
> ... Ivy ... pc ...
Mea culpa, mea culpa, mea maxima culpa. I have never tested ivy against
"error situations". I took the language as it was and enhanced it with
additional acquisition-hardware-related "functions" repairing some
problems that I had found. Inside of these functions if anything,
including actual parameters, was wrong the setjump/longjump ( if I
remember well ) was used ( as also used in the "standard" ivy runtime
library ). In your previous mail you say that if one was to take Ivy, one
would need to change syntax ( "operators" ), add features ( "arrays" ),
... - this would mean that one is creating a new language ( similar to
Ivy, but not Ivy ). It's not worth the time. I once choose Ivy because it
was extemely simple and sufficient for my purposes - I just needed to add
some "built-in" functions.
> ...
Finally, I vote for S-Lang, and if it's too simple for CINT.
( Of course, I consider the Ivy option still open. )
Jacek.

Reply via email to