Re: SIMPLE: A language-independent tree IR

[Diego Novillo <dnovillo at redhat dot com>]

On Wed, 02 Jan 2002, Joe Buck wrote:

> > There may be other C assumptions in the current tree forms to consider and
> > maybe change - for example, are the C rules that overflow of signed
> > integer arithmetic is undefined appropriate for all languages, or is this
> > another matter that should be explicitly represented in the various forms?
> 
> Perhaps this is best handled in a framework like SIMPLE by using different
> OPs.  That is, if lang-1 says that overflow of signed integer arithmetic
> is undefined but lang-2 requires an exception to be thrown, then a
> different OP would be used; alternatively, the latter could be defined in
> terms of the former plus extra code.
> 
Yes.  My initial plan is to model the different tree -> SIMPLE
converters after the existing tree -> RTL ones.  All the
language semantics are encoded in the current expanders, so it
should be a matter of making the expanders output SIMPLE trees
instead of RTL.

I'm not actually sure how true the above is.  We may have some
semantics encoded in other places.  Once the semantics have been
explicitly expanded into SIMPLE, then we can have a single
SIMPLE->RTL expander.


Diego.