This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re[2]: RTL/tree (from gcc) and Scheme


Hello Per,

Sunday, August 10, 2003, 8:51:26 AM, you wrote:

>> Has kawa the same problem?

PB> You asked for a GCC front-end for Scheme.

no, I've asked not for GCC front end, but for GCC's front end.
It means RTL/tree.

PB> Now you're asking for being able "to compile function in a run-time
PB> up to native code."  That's a completely unrelated problem with a
PB> different set of issues.

compilation in run-time is one of key issues in languages like CL.
Sorry, for any misleading formulations, my motivations are expressed
here: http://www.cliki.net/CLVM and
http://www.cliki.net/Valery%20Khamenya 

PB> I can think of two solutions if using
PB> Kawa:
PB> * Kawa (I don't know about Bigloo) will compile to bytecode at run-time.
PB> A modern Java VM will Just-in-Time (JIT) compile the bytecode the native
PB> code in-time fly.

nice, could I treat your answer that kawa *already* can compile in
run-time?

PB> * You can compile the bytecode, save them in a file, invoke Gcj to
PB> create a DLL, and load that at run-time.  At least in theory and with
PB> a little bit of work - I don't know if anyone has done it.

solution through a file is not really nice. Indeed, is it then worth
of single function compilation?

>> why?..

PB> Because your question doesn't make sense.  RTL and Tree are data
PB> structures internal to the GCC compiler.  They are not designed
PB> for anything else.

hm, I am a bit confused. At least Fortran, Ada and Java provided with
gcc distribution use the same RTL/tree as far as I understood... or am
I wrong?.. 

-- 
Best regards,
Valery                            mailto:khamenya@mail.ru


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]