• "Ben Hinkle" <bhinkle4 juno.com> Jan 02 2004
• Ilya Minkov <minkov cs.tum.edu> Jan 02 2004
• Stephan Wienczny <wienczny web.de> Jan 03 2004
• "Ben Hinkle" <bhinkle4 juno.com> Jan 03 2004
• "Walter" <walter digitalmars.com> Jan 03 2004
• "Ben Hinkle" <bhinkle4 juno.com> Jan 03 2004
• "Walter" <walter digitalmars.com> Jan 03 2004

Since it's the new year and we all have a few days off I figure'd I'd try to
see just how bad it would be to modify the C++ front end to compile D.

After one day of search/replace cxx (and cp and tons of other stuff) with
dmd and one parser change to recognize "extern (C)" (in parse.y and the
wonderfully named file "spew.c") I got hello world to compile and run
(linked with libstc++):

extern (C) void printf(char*, ...);
extern (C) int main(int argc, char**argv)
{
    printf("hello world.\n");
    return 0;
}

[/home/bhinkle/d/gcc/build]845 %
../gcc-3.3.2/configure --enable-languages=c,dmd
[snip]
[/home/bhinkle/d/gcc/build]846 % make bootstrap
[snip]
[/home/bhinkle/d/gcc/build]847 % cd gcc
[/home/bhinkle/d/gcc/build/gcc]848 % gdmd
/home/bhinkle/hello.d -L. -L/home/bhinkle
[/home/bhinkle/d/gcc/build/gcc]849 % a.out
hello world.

Now my goal (something that would actually require work) is to get phobos
compiled. It doesn't look like there are any templates in there so templates
will wait. I figure I'll run into a brick wall at some point but what the
heck.

Anyway, I'm not sure how far I'll get but happy new year.

-Ben

Ben Hinkle wrote:
 Since it's the new year and we all have a few days off I figure'd I'd try to
 see just how bad it would be to modify the C++ front end to compile D.
 
 After one day of search/replace cxx (and cp and tons of other stuff) with
 dmd and one parser change to recognize "extern (C)" (in parse.y and the
 wonderfully named file "spew.c") I got hello world to compile and run
 (linked with libstc++):


Woohoo! So you're giving it a try another way around! You shall need a 
lot of good luck in rewriting the C++ front-end, if this really is what 
you intend to do...

I was not going to dig within G++ until i understand and can dump DMD... 
But i had realized a while ago that i shall for a long while have time 
for neither. I'm sorry.

Good luck and happy new year.

-eye

Ben Hinkle wrote:
 Since it's the new year and we all have a few days off I figure'd I'd try to
 see just how bad it would be to modify the C++ front end to compile D.
 
 After one day of search/replace cxx (and cp and tons of other stuff) with
 dmd and one parser change to recognize "extern (C)" (in parse.y and the
 wonderfully named file "spew.c") I got hello world to compile and run
 (linked with libstc++):
 
 extern (C) void printf(char*, ...);
 extern (C) int main(int argc, char**argv)
 {
     printf("hello world.\n");
     return 0;
 }
 
 [/home/bhinkle/d/gcc/build]845 %
 ../gcc-3.3.2/configure --enable-languages=c,dmd
 [snip]
 [/home/bhinkle/d/gcc/build]846 % make bootstrap
 [snip]
 [/home/bhinkle/d/gcc/build]847 % cd gcc
 [/home/bhinkle/d/gcc/build/gcc]848 % gdmd
 /home/bhinkle/hello.d -L. -L/home/bhinkle
 [/home/bhinkle/d/gcc/build/gcc]849 % a.out
 hello world.
 
 Now my goal (something that would actually require work) is to get phobos
 compiled. It doesn't look like there are any templates in there so templates
 will wait. I figure I'll run into a brick wall at some point but what the
 heck.
 
 Anyway, I'm not sure how far I'll get but happy new year.
 
 -Ben
 
 
 


Do you have a website?
Sourceforge project etc?

Stephan

 You did not use the 3.4 branch?!?


umm. whoops. I just got the latest release a few weeks ago or so. I'll check
it out.

 Do you have a website?
 Sourceforge project etc?


if I actually get anywhere with it

 Stephan

"Ben Hinkle" <bhinkle4 juno.com> wrote in message
news:bt3thn$2i0a$1 digitaldaemon.com...
 Since it's the new year and we all have a few days off I figure'd I'd try


 see just how bad it would be to modify the C++ front end to compile D.


That is an interesting approach. It just might work!

 Now my goal (something that would actually require work) is to get phobos
 compiled. It doesn't look like there are any templates in there so


 will wait. I figure I'll run into a brick wall at some point but what the
 heck.


D templates are easy - it's entirely handled within the front-end.

"Walter" <walter digitalmars.com> wrote in message
news:bt7qeh$294u$2 digitaldaemon.com...
 "Ben Hinkle" <bhinkle4 juno.com> wrote in message
 news:bt3thn$2i0a$1 digitaldaemon.com...
 Since it's the new year and we all have a few days off I figure'd I'd




 to
 see just how bad it would be to modify the C++ front end to compile D.


 That is an interesting approach. It just might work!


I added alias and uchar, ushort and friends with no problem but my next goal
was dynamic arrays and that has been harder. I haven't figured out enough
about the G++ front end to make any real progress on that. It turns out the
C notion of arrays is pretty hard-wired into gcc. Also I started ripping out
the preprocessor but that caused havoc in the parse and lex code (and that
lovely spew.c). Right now I'm looking at plugging in your lexer since it has
the least dependencies on anything else and is *much* simpler than the G++
cpp-based lexer. I'm starting to think my approach will end up as a rather
extensive transplant of certain G++ chunks with the DMD equivalents.

 Now my goal (something that would actually require work) is to get




 compiled. It doesn't look like there are any templates in there so


 will wait. I figure I'll run into a brick wall at some point but what




 heck.


 D templates are easy - it's entirely handled within the front-end.


easy, eh?  Don't get my hopes up ;-)

"Ben Hinkle" <bhinkle4 juno.com> wrote in message
news:bt84h2$2n84$1 digitaldaemon.com...
 Also I started ripping out
 the preprocessor but that caused havoc in the parse and lex code (and that
 lovely spew.c).


spew, gotta love it!

 Right now I'm looking at plugging in your lexer since it has
 the least dependencies on anything else and is *much* simpler than the G++
 cpp-based lexer.


I had the advantage here because whenever the D lexer started looking
complicated, I simplified the language. Suggest simplifying the C++ language
to make scanning easier, and one's compiler cajones get questioned <g>.

 D templates are easy - it's entirely handled within the front-end.




Once you get the rest of the DMD front end in, it should be easy!