--- Begin Message ---
Subject: |
29.1; cl-defstruct'ed struct errors out in the constructor compiler macro if a slot is called "quote" |
Date: |
Tue, 9 Jan 2024 21:32:06 +0900 |
* The problem
If a struct is defined to have a slot called "quote", the constructor
will fail during the compiler macro phase.
(require 'cl-lib)
(cl-defstruct my-test quote)
(make-my-test :quote "'") ; error: (invalid-function "'")
(put 'make-my-test 'compiler-macro nil) ; disable the compiler macro
(make-my-test :quote "'") ; => #s(test "'")
I have tested Emacs 25.3, 26.3, 27.2, 28.2, and 29.1; this error is in
all these versions.
* Why this is a bug
This feels like a bug to me, because it's not clear that this would be
an unsupported use case, and because of the inconsistent behavior
between the constructor being inlined vs. without a compiler macro.
* Potential directions
This could be "fixed" by either digging into cl-defstruct and cl-lib
then fixing it properly, or by simply declaring in the documentation
that this is not supported. I wish for a proper fix, but the
documentation workaround also seems fine to me.
* Additional info
This is what the compiler macro for make-my-test expands into:
(cl-block make-my-test
(record
("'" my-test)
"'"))
In GNU Emacs 29.1 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.38,
cairo version 1.17.8)
Windowing system distributor 'The X.Org Foundation', version 11.0.12101010
System Description: Arch Linux
--- End Message ---
--- Begin Message ---
Subject: |
Re: bug#68345: 29.1; cl-defstruct'ed struct errors out in the constructor compiler macro if a slot is called "quote" |
Date: |
Wed, 10 Jan 2024 14:55:01 +0200 |
> Cc: 68345@debbugs.gnu.org
> Date: Wed, 10 Jan 2024 07:46:56 +0900
> From: Kisaragi Hiu via "Bug reports for GNU Emacs,
> the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
>
> On 2024/01/09 23:59, Gerd Möllmann wrote:
> > Could you please check with Emacs/master? Seems to work there.
>
> It works as I'd expect at master (commit
> fccaeabc959f5403ce49744030bd2620352b59f8), without errors. I should've
> checked master before filing this report - sorry about that.
>
> I might run a bisect later to see how it was fixed, but this appears to
> no longer be an issue.
Thanks, closing.
--- End Message ---