I think I provided a reference the lady time this topic came up. On my
phone, so apologes if these don't turn out. It's one of these (they
may actually be the same -- had trouble viewing on my phone):
http://www.cs.cornell.edu/talc/papers/compile_rtcg-tr.pdf
http://www.cs.washington.edu/homes/djg/papers/compile_rtcg.pdf
The other possibility is to forbid dynamic linking that would require
codegen. Not ideal, but workable.
Sandro
On 2011-04-01, at 18:16, "Jonathan S. Shapiro" <[email protected]> wrote:
On Thu, Mar 31, 2011 at 8:31 AM, Sandro Magi <[email protected]
> wrote:
Depends what you mean by "static compilation". Cyclone had a paper on
dynamic code generation by stitching together code templates.
Can you provide a pointer to that paper?
You can
support link time code generation using this technique using all
native
code.
In spite of how I may sound, I'm actually a big fan of link-time
code generation. But when link-time code generation is combined with
dynamic libraries, you've crossed the line into runtime code
generation. This places a significant burden on the target
environment, and a significant burden of complexity (therefore
assurance) on the runtime environment.
It's the burden of runtime assurance that mainly concerns me here.
shap
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev
