[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DotGNU] compat

From: Gopal V
Subject: Re: [DotGNU] compat
Date: Sun, 12 Jan 2003 21:18:32 +0530
User-agent: Mutt/1.2.5i

If memory serves me right, Patrick Pletscher wrote:
> Yes, the compiler compiles everything and after that he starts the ILasm,
> but
> the author of the compiler will perhaps replace the 2 lines, where this
> procedure is used...

Process stuff might not be hard to implement .... I was just keeping off
internal calls for the 0.5.0 release period ...

> somebody wants to compile it itself.

  newobj void [mscorlib]System.String::.ctor(char[]) 

Bingo ! ... it should be 
  newobj instance void [mscorlib]System.String::.ctor(char[])

I think I've fixed it up in the assembler...

The problem here is that our "super-fast" assembler was intended for
compiler writers ... so it sort of assumes a lot of things have already
been figured out ...

For example ,

  callvirt char [] [mscorlib]System.String::ToCharArray () 

will be interpreted as a function without any "static" or "instance" flags .
(It's fixable, but I'm into some other stuff right now).

We sort of use it as a read & assemble program instead of verifying that all 
the linkages are correct. We need this behaviour to power our C compiler.

So it would assemble all the seperate C programs into IL , assemble each into
.obj file and link them together ... 

This is one of the secrets of why our compiler is faaast (now all remains is 
rhys to fixup on-demand loading of classes for the compiler and it will be 
on fire !)...

> would be really cool, if I could code the programs in VIM and after that
> compile it on the shell....

well I already do that :) ... 

The difference between insanity and genius is measured by success

reply via email to

[Prev in Thread] Current Thread [Next in Thread]