Re: Emacs design and architecture. How about copy-on-write?

[Ihor Radchenko]

Eli Zaretskii <eliz@gnu.org> writes:

>> My solution (which I've put into practice in redisplay) is to save those
>> values before sensitive code is executed, and to refer to those saved
>> values within said code.
>
> That is _everyone's_ solution, not just yours.  But it is not as easy
> in practice as it may sound.  E.g., imagine a subroutine that is
> called by some higher-level functions, where both the callers and the
> subroutine need to access the same variable.  When other threads are
> running, there's no longer a guarantee that both the caller and the
> callee will see the same value of that variable.  If they must use the
> same value, you now need to pass that variable to the callee via its
> API, and this is not scalable when you have more than a couple,
> especially if the callee is not called directly, but via several
> intermediate callers.

May the state be captured and passed to the callee under the hood?
Something like forking, but using thread-local state.

For lexical bindings, it is just a matter of passing over current
Vinternal_interpreter_environment; and for global bindings, it is a
matter of passing whatever is altered by the caller thread (with
copy-on-write, this info should be available).

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>