On Wed, Oct 30, 2013 at 10:26 PM, Andrew MacLeod wrote:
> I've made 4 attempts now to split gimple.[ch] into reasonable component
> parts, and I've finally found something that I can make work and fits my
> plans.
>
> I've attached a diagram to (hopefully :-) clarify things.
That's not a class hi
On 11/01/2013 10:45 AM, Diego Novillo wrote:
On Wed, Oct 30, 2013 at 10:26 PM, Andrew MacLeod wrote:
I've made 4 attempts now to split gimple.[ch] into reasonable component
parts, and I've finally found something that I can make work and fits my
plans.
I've attached a diagram to (hopefully :-)
On Fri, Nov 1, 2013 at 10:56 AM, Andrew MacLeod wrote:
> There are no classes. It's an include dependency relationship of the files
> I'm trying to split out Ie who includes/has visibility to a
> file/component.
Ah, thanks.
> Isnt "gimplify" clear to everyone? :-) I've even been referring
On 11/01/2013 10:59 AM, Diego Novillo wrote:
On Fri, Nov 1, 2013 at 10:56 AM, Andrew MacLeod wrote:
Isnt "gimplify" clear to everyone? :-) I've even been referring to an
interface to the backend target info for the front end as "targetify" lately
:-)
Yes, but only when we are coming from th
On Fri, Nov 1, 2013 at 11:11 AM, Andrew MacLeod wrote:
> Longer term... the BE will not be calling into gimplify at all.. it will be
> purely a front end thing. As direct tree access vanishes, the optimizers
> will be forced to generate gimple directly, as they should.
Right. Which is the rout
On 11/01/2013 11:17 AM, Diego Novillo wrote:
On Fri, Nov 1, 2013 at 11:11 AM, Andrew MacLeod wrote:
Longer term... the BE will not be calling into gimplify at all.. it will be
purely a front end thing. As direct tree access vanishes, the optimizers
will be forced to generate gimple directly,
On Fri, Nov 1, 2013 at 11:23 AM, Andrew MacLeod wrote:
> gimplify would be routines that work on trees since that's what the front
> end will present. And thats why the backend will no longer need to gimplify
> anything... It'll just build stuff using the gimple-stmt build and the
> gimple type
