bug#46326: 27.1.50; Excessive memory allocations with minibuffer-with-setup-hook

[mail@daniel-mendler.de]

On 2021-02-08 10:25, jakanakaevangeli wrote:
> I seem to have solved the problem by doing the following simple
> substitutions in add-hook and remove-hook:
> 
>         (alist-get ... #'equal) -> (alist-get ...)
>         equal -> eq
>         member -> memq
>         delete -> delq
> 
> This changes the behaviour of add/remove-hook a little, for example
> 
>         (add-hook    (lambda () (test)))
>         (remove-hook (lambda () (test)))
> 
> will not be reverse operations anymore, since the two lambdas are not
> eq
> as they are created separately. But the performance gain for big
> non-compiled closures might be worth considering.

Thank you, this also seems like a possible solution. However I fear that
it
is way to intrusive and could break a lot of existing code. As I
mentioned
before set-transient-map avoids adding a letrec-lambda via add-hook
because
it would make problems with equal. This means the "broken nature" of
add-hook
is somehow accepted pervasively over the Emacs codebase. My fix is less
intrusive
in the sense that it would just fix `minibuffer-with-setup-hook'.