Re: [RFA][2/5] New port: Cell BE SPU (valops.c fix)

["Ulrich Weigand" <uweigand@de.ibm.com>]

Jim Blandy wrote:

> The key is that r2v knows how to both decode the register's contents,
> and how to re-encode both the full contents, or a portion of the
> contents, so it can leave enough information in the closure to tell
> v2r how to handle component writes.

I see.  So basically, instead of encoding the original type in the
struct value, you have accessor functions in the struct value, and
the back-end can install different versions of those functions if
it needs to handle types differently.  That would work as well.

I've prepared a set of two patches that fix my spu problem in a way
that I feel would be a step into the direction you describe.  I'll
post those shortly ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com