Re: possibly spurious bytecomp warning:

[Michael Heerdegen]

Robin Tarsiger <rtt@dasyatidae.com> writes:

>   (cl-loop
>    for c in
>    '(?I ?o)
>    do
>    (when (assoc  c eww-link-keymap)
>      (delete (assoc  c eww-link-keymap) eww-link-keymap))) ; <--
>
> Here, the list being modified is a keymap, so the first element
> is the keymap type symbol and assoc never returns it. And notably,
> delete's _docstring_ doesn't guarantee in-place modification for
> non-head deletions, but the Info node for delq/delete describes both
> functions as doing cdr splicing in that case, so assuming that's meant
> as normative, this will operate correctly.

I think this is correct.

Maybe using higher-level functions for keymaps (e.g. `lookup-key') would
be better.

> It probably wouldn't be reasonable for bytecomp to try to distinguish
> this, though...

I would not harm if it could, but how would it do that?

Michael.