bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

[Drew Adams]

How did the signature of `completing-read' get changed?  I didn't notice
any proposal or discussion about this in emacs-devel@gnu.org.  Did I
just miss it somehow?

It used to be that _any_ REQUIRE-MATCH value that is not `t', nil,
`confirm', or `confirm-after-completion' behaves like `t', except that
type RET doesn't exit if what you type does non-null completion.

That's no longer true if the value is a function!  This completely
changes the behavior of `completing-read'.

Not happy with the result, and not happy with how the process - how this
was done, if it wasn't discussed openly in emacs-devel.


In GNU Emacs 29.1 (build 2, x86_64-w64-mingw32) of 2023-08-02 built on
 AVALON
Windowing system distributor 'Microsoft Corp.', version 10.0.19045
System Description: Microsoft Windows 10 Pro (v10.0.2009.19045.3448)

Configured using:
 'configure --with-modules --without-dbus --with-native-compilation=aot
 --without-compress-install --with-tree-sitter CFLAGS=-O2'

Configured features:
ACL GIF GMP GNUTLS HARFBUZZ JPEG JSON LCMS2 LIBXML2 MODULES NATIVE_COMP
NOTIFY W32NOTIFY PDUMPER PNG RSVG SOUND SQLITE3 THREADS TIFF
TOOLKIT_SCROLL_BARS TREE_SITTER WEBP XPM ZLIB

(NATIVE_COMP present but libgccjit not available)

bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

[Stephen Berman]

On Tue, 3 Oct 2023 22:11:41 +0000 Drew Adams <drew.adams@oracle.com> wrote:

> How did the signature of `completing-read' get changed?  I didn't notice
> any proposal or discussion about this in emacs-devel@gnu.org.  Did I
> just miss it somehow?
>
> It used to be that _any_ REQUIRE-MATCH value that is not `t', nil,
> `confirm', or `confirm-after-completion' behaves like `t', except that
> type RET doesn't exit if what you type does non-null completion.
>
> That's no longer true if the value is a function!  This completely
> changes the behavior of `completing-read'.
>
> Not happy with the result, and not happy with how the process - how this
> was done, if it wasn't discussed openly in emacs-devel.

There was a short discussion, after the change was made, starting here:
https://lists.gnu.org/archive/html/emacs-devel/2022-06/msg00539.html

Steve Berman

bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

[Drew Adams]

> > How did the signature of `completing-read' get changed?
> > I didn't notice any proposal or discussion about this
> > in emacs-devel@gnu.org.  Did I just miss it somehow?
> >
> > It used to be that _any_ REQUIRE-MATCH value that is
> > not `t', nil, `confirm', or `confirm-after-completion'
> > behaves like `t', except that type RET doesn't exit if
> > what you type does non-null completion.
> >
> > That's no longer true if the value is a function!
> > This completely changes the behavior of `completing-read'.
> >
> > Not happy with the result, and not happy with how the
> > process - how this was done, if it wasn't discussed
> > openly in emacs-devel.
> 
> There was a short discussion, after the change was made, starting here:
> https://urldefense.com/v3/__https://lists.gnu.org/archive/html/emacs-
> devel/2022-
> 06/msg00539.html__;!!ACWV5N9M2RV99hQ!Pi4vEIugzynWXlOXCj_8GVnUyeP_8Q9i9ysV
> ZwoUAmd2dc4qwMRUMS8Ce9W_d_8GAlmYBaDccZg8x2-utGVJed4B$

I see; thank you!

Yes, very ugly.  And no proposal or discussion;
just Lars changing things.  At least Stefan
spoke up (though not about the basic breaking
of compatibility) - after the fait accompli.

I suppose I should have guessed it was something
like that.  Wish I'd have seen it at the time,
and realized what the overall effect is.

Really too bad.

The justification given: "adding a new parameter
for this use case seemed a bit overboard."  So
just break what that argument has always been
about, and reuse it for something altogether
different?  Sigh.

bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

[Eli Zaretskii]

> From: Drew Adams <drew.adams@oracle.com>
> Date: Tue, 3 Oct 2023 22:11:41 +0000
> 
> How did the signature of `completing-read' get changed?  I didn't notice
> any proposal or discussion about this in emacs-devel@gnu.org.  Did I
> just miss it somehow?

It was discussed here:

  https://lists.gnu.org/archive/html/emacs-devel/2022-06/msg00539.html

bug#66328: 29.1; Incompatible change to `completing-read' breaks existing code

[Eli Zaretskii]

> Cc: "66328@debbugs.gnu.org" <66328@debbugs.gnu.org>
> From: Drew Adams <drew.adams@oracle.com>
> Date: Wed, 4 Oct 2023 02:04:01 +0000
> 
> > > How did the signature of `completing-read' get changed?
> > > I didn't notice any proposal or discussion about this
> > > in emacs-devel@gnu.org.  Did I just miss it somehow?
> > >
> > > It used to be that _any_ REQUIRE-MATCH value that is
> > > not `t', nil, `confirm', or `confirm-after-completion'
> > > behaves like `t', except that type RET doesn't exit if
> > > what you type does non-null completion.
> > >
> > > That's no longer true if the value is a function!
> > > This completely changes the behavior of `completing-read'.
> > >
> > > Not happy with the result, and not happy with how the
> > > process - how this was done, if it wasn't discussed
> > > openly in emacs-devel.
> > 
> > There was a short discussion, after the change was made, starting here:
> > https://urldefense.com/v3/__https://lists.gnu.org/archive/html/emacs-
> > devel/2022-
> > 06/msg00539.html__;!!ACWV5N9M2RV99hQ!Pi4vEIugzynWXlOXCj_8GVnUyeP_8Q9i9ysV
> > ZwoUAmd2dc4qwMRUMS8Ce9W_d_8GAlmYBaDccZg8x2-utGVJed4B$
> 
> I see; thank you!
> 
> Yes, very ugly.  And no proposal or discussion;
> just Lars changing things.  At least Stefan
> spoke up (though not about the basic breaking
> of compatibility) - after the fait accompli.
> 
> I suppose I should have guessed it was something
> like that.  Wish I'd have seen it at the time,
> and realized what the overall effect is.
> 
> Really too bad.
> 
> The justification given: "adding a new parameter
> for this use case seemed a bit overboard."  So
> just break what that argument has always been
> about, and reuse it for something altogether
> different?  Sigh.

I see no reason to revert that change, so I'm closing this bug.