Re: [PATCH] [WIP] Port feature/native-comp to Windows.

[Andrea Corallo]

Eli Zaretskii <address@hidden> writes:

>> From: Andrea Corallo <address@hidden>
>> Date: Sat, 09 May 2020 16:27:18 +0000
>> Cc: "address@hidden" <address@hidden>
>> 
>> Nicolas Bértolo <address@hidden> writes:
>> 
>> > PS: I don't understand why the freloc table is necessary. Is it impossible 
>> > to
>> > rely on the dynamic linker to solve that when calling dlopen()?
>> 
>> The problem is that being Emacs an executable we do not export the
>> symbols to be used by the dynamic linker.
>
> You mean, the symbols of primitives that compiled Elisp code calls?

Yes

>> I think it should be possible (depending on the tool-chain) but I never
>> tried.  What I'm not sure of is how much is it portable given is not a
>> common scenario.  Because typically afterwards the answer comes up to be
>> "not enough" at the time I decided to handle it inside, the generated
>> code should be similar.
>
> But that is even less portable, isn't it?  Or am I missing something?

No should be portable, the mechanism is very simple, we have just some
memory in the .eln (static variables) that we fill with the right
pointers during load.

Andrea

-- 
address@hidden