Re: Emacs Lisp JIT Compiler

[Ergus <spacibba@aol.com>]

Hi!

Sorry for the bother, but I am very interested in this kind of features!

I have 2 simple yes/not questions:

1) Any progress in the JIT work?

2) Do you think we will see it soon in the next release or year for example?

And an extra if you have some more time to reply.

3) I saw that the option to translate Lisp code into C and compile it is
not considered; but I couldn't understand why and how is it better to
use libJIT than a native compiler+binutils. Lisp-c or equivalents could
work to distribute all the emacs internal functions already compiled but
maybe the advanced users could enable it as an experimental feature for
his own functions (with some configuration). Also consider the
pedagogical value of this translator's code and the potential use cases.

Best,
Ergus

On Sat, Aug 25, 2018 at 09:51:02AM +0300, Eli Zaretskii wrote:
>> From: Tom Tromey <tom@tromey.com>
>> Cc: Tom Tromey <tom@tromey.com>,  emacs-devel@gnu.org
>> Date: Fri, 24 Aug 2018 15:03:21 -0600
>>
>> >>>>> "Eli" == Eli Zaretskii <eliz@gnu.org> writes:
>>
>> Eli> Maybe I'm missing something, but I didn't see how to do that with just
>> Eli> a type definition.  jit_value_create_long_constant is special in that
>> Eli> it allows to create 64-bit constants in a 32-bit build, by allocating
>> Eli> a 64-bit buffer and storing it address in the jit_value_t object it
>> Eli> returns.
>>
>> Ah, I didn't pick up on that.  My illness makes me somewhat mentally
>> fuzzy.
>
>It's not just your illness, it's also the libjit docs, which never say
>that.  I learned that by reading the libjit sources.
>
>> Anyway maybe a helper function could be used so there's just a single #if.
>
>Yes, I think so.  I will see what can be done about it.
>