> In the context of ELisp, for example, we wouldn't an incorrect type
declaration to risk crashing your editor, which limits significantly
what can be done and how we can do it.

I do not get your point. Errors are a horrible problem we face in every part of all the code we run. Software that catches errors is a necessity, but it does not work perfectly. I get blown away many times a day and have to reboot my Chromebook. I cannot easily list all of the things I am constantly watching out for: network service disconnection, spilling my drink on my keyboard, power failures ..., relatives dying, doctors appointments, ...

I regard errors from declarations as wonderful reports that usually show up during SBCL compilation.

Bob

On Mon, Feb 19, 2024 at 8:51 PM Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> However, I have been hacking Lisp hard since 1967, and to me the assertion
> that a Lisp must support THE and DECLARE is about as obvious as the
> assertion that doctors must wash their hands. An untold number of infants'

There are several subtly different ways to interpret that assertion,
actually, when you're working on the actual implementation of
a compiler.

In the context of ELisp, for example, we wouldn't an incorrect type
declaration to risk crashing your editor, which limits significantly
what can be done and how we can do it.

Stefan