Why so many theoretical discussions when ...

[Brad Roberts]

Nick Sabalausky wrote:
> "Walter Bright" <newshound1 at digitalmars.com> wrote in message 
> news:g9a7eg$1k9b$3 at digitalmars.com...
>> BCS wrote:
>>> We agree on many things but when we can't change DMD all we have left is 
>>> to argue over the details (some of us think this is "fun")
>> You can always change gdc, which is based on dmd, to try things out.
> 
> Mucking around with GCC is an absolute mess (doubly true on windows). Not 
> that I'm disagreeing with you, though. FWIW, a D compiler written in a 
> better language (like D!) would be great for trying things out, but I 
> suppose that's just stating the obvious ;)  (BTW, What is LLVM written in? 
> C++, I assume? Haven't really had a chance to look into it yet.) 
> 

Working on the dmd frontend within gdc isn't that bad at all.  The
instructions on how to compile gdc were understandable last time I tried
it (though admitedly I'd built gcc on more than one occasion prior to
being introduced to d).  From there, for the vast majority of any
changes, you wouldn't even have to look at the gcc or gcc -> dmd glue
code.  You could look exclusively at the dmd frontend.

Really, it's inertia that keeps more people from playing with the
compiler than the actually difficulty of compiling it.

Later,
Brad