Re: Is there an Euphoria compiler?
>Is there, and if not, will there ever be, an Euphoria compiler? It seems
>to me that tacking on the whole interpreter to a program is not efficient.
Not efficient ??
Wanne bet, a binded version of TextGUI is smaller than the MSDOS Editor.
which is an com file that launches qbasic /ascii_mode or something like
that).. so not the com file that comes wiht DOS, but the qbasic.exe or the
editor found with Win95.
My point is, an quake clone in Euphoria, would be slow, but so much more
smaller than the real quake.
Euphoria programs have binded a minimum about 160 kb, but from that point on
it increases almost nothing. Even the biggest programs only a few Kb. Big
programs, were size *does* count (unlike small programs) are still
relatively small and easy to distribute. Plus, remember, most of the 160 kb
is put in the extended memory thanks to the memory extender.
And last, but not least, it could be compiled much.
As an interpreter, Euphoria can do certain stuff, that compiled would be a
call to a piece of code, just like now.
Example, a decleration as object, you can't compile *anything* since you do
not know which data type it is gonna hold. You would create a mixture of
the interpreters code and a bit of code to call the interpreters code. What
exactly is the difference with a binded version ?? NONE
Yes, the code is now seperated and we needed to compile. Plus, compiling
would mean, Euphoria's safety would be gone.
Nope, from an Euphoria source code, there is no way to compile it into
anything significantly faster, since the compiler wouldn't have enough
information, that Euphoria does not need to have.
It would only need a lot more memory, and would run a lot less stable.
Compiled languages are fast, not because they compile (they wish), but
because they can optimize by asking the right questions. Asking such
question would eliminate Euphoria elegance.
Ralf Nieuwenhuijsen
nieuwen at xs4all.nl
|
Not Categorized, Please Help
|
|
