IMO supercompilation is related to abstract interpretation. In fact an
abstract interpreter can behave as a concrete interpreter until multiple
program states merge into a single state. In that case the interpreter has
to give up precision and introduce abstract values. This technique is
called abs
The external STG interpreter implements the RTS semantics and features, so
if we apply the calculating correct compiler method to the external STG
interpreter code then we should get an IR that will include the RTS code
also.
On Wed, Jun 9, 2021 at 12:21 AM Carter Schonwald
wrote:
> How would th
How would this be used to generate the rts automatically? I’m intrigued /
would like to understand what you’re envisioning design wise for that leg.
On Tue, Jun 8, 2021 at 5:34 PM Csaba Hruska wrote:
> Cmm is too low level, I've implemented the primops in haskell in a high
> level way, including
Cmm is too low level, I've implemented the primops in haskell in a high
level way, including the out of line primops with the rts related parts
(scheduler, io manager).
see:
https://github.com/grin-compiler/ghc-whole-program-compiler-project/blob/master/external-stg-interpreter/lib/Stg/Interpreter/
The stm impl In ghcjs might be a helpful comparative example on that front.
Though I guess more broadly does this necessitate having a model of the Cmm
semantics for the out of line primops ?
On Tue, Jun 8, 2021 at 5:10 AM Simon Peyton Jones via ghc-devs <
ghc-devs@haskell.org> wrote:
> I wonder
I wonder if there was an attempt in the past to create an abstract interpreter
for GHC Core or STG to approximate the program runtime behaviour?
No, not that I know of. Because of all the primops, concurrency, STM, etc, it
would be something of a challenge. The AAM story could be interesting…
Yes, I've read Thomas Schilling's PhD thesis (Trace-based
Just-in-timeCompilation for Lazy Functional Programming Languages) a couple
of times.
My STG interpreter almost supports all kinds of primops that GHC does, and
I plan to add the missing ones in the future.
I'd like to use literally the sam
Correction it was Thomas schilling !!
And you can Google his Phd thesis trace based just in time compilation for
lazy functional programming languages
And the associated code is on his GitHub nominolo/ lambdachine though I
think it was last touched 7 years ago
On Mon, Jun 7, 2021 at 11:54 AM Cart
I’m not aware of any currently.
I would be curious about the now relatively old work that Max bolingbroke
did for his PhD (I think it was sortah a ghc to Lua JIT?!?)
An important question is : what questions do you want the abstract
interpreter to suport?
On Mon, Jun 7, 2021 at 10:19 AM Csaba Hr
