Re: Introducing Jitgen

[Ludovic Courtès]

Hello!

Better late than never...

Noah Lavine <address@hidden> writes:

[...]

> The goal of Jitgen is to generate both of these from the following 
> instructions:
>
> '((ip <- ip + ipup)
>   (tmp1 <- fp - spup)
>   (tmp2 <- sp - tmp1)
>   (tmp3 <- tmp2 == (n_val & 0x7))
>   (branch-if tmp3 good_path)
>   (store-relative ipp 0 ip)
>   (store-relative spp 0 sp)
>   (store-relative fpp 0 fp)
>   (return jit_return_wrong_num_args)
>   (label good_path)
>   (sp <- sp + spup)
>   (store-relative sp 0 SCM_UNDEFINED)
>   )

Excellent!

The challenge will be to see if it’s sufficiently expressive to express
all the actual VM instructions, and allows the generated code to be
about the same as the hand-written one.

> My biggest question is, is this something that you would be interested
> in having in Guile?

I think so, yes.  Not in 2.0, I guess, but when it starts stabilizing it
could go into a branch.

> If so, then also what do you think of the code style and interfaces,
> and do you have any other feedback?

The code style looks OK to me.

This approach raises bootstrapping issues, or semi-issues.  For
instance, tarballs could come with all the pre-generated C files
containing the VM instructions, but we may also have to check them in.
That’s how psyntax bootstrapping is handled already, so it’s probably
OK.

Thanks,
Ludo’.