Re: master e4df6203e2 3/3: More robust `gnu` pattern (bug#56249)

Re: master e4df6203e2 3/3: More robust `gnu` pattern (bug#56249)

[Stefan Monnier]

>     More robust `gnu` pattern (bug#56249)
>     * lisp/progmodes/compile.el (compilation-error-regexp-alist-alist):
>     Don't allow spaces in the file names, to avoid false matches.

Isn't this a regression?


        Stefan

Re: master e4df6203e2 3/3: More robust `gnu` pattern (bug#56249)

[Mattias Engdegård]

29 juni 2022 kl. 19.25 skrev Stefan Monnier <monnier@iro.umontreal.ca>:

>>    Don't allow spaces in the file names, to avoid false matches.
> 
> Isn't this a regression?

Sorry, that should have been "Don't allow spaces in the beginning of file names".

Re: master e4df6203e2 3/3: More robust `gnu` pattern (bug#56249)

[Stefan Monnier]

Mattias Engdegård [2022-06-29 20:15:42] wrote:
> 29 juni 2022 kl. 19.25 skrev Stefan Monnier <monnier@iro.umontreal.ca>:
>>>    Don't allow spaces in the file names, to avoid false matches.
>> Isn't this a regression?
> Sorry, that should have been "Don't allow spaces in the beginning of file names".

Ah... makes more sense, thanks,


        Stefan

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Michael Heerdegen]

Mattias Engdegård <mattiase@acm.org> writes:

> --- a/lisp/emacs-lisp/rx.el
> +++ b/lisp/emacs-lisp/rx.el
> @@ -1110,6 +1110,14 @@ can expand to any number of values."
>    (append rx--builtin-forms rx--builtin-symbols)
>    "List of built-in rx names.  These cannot be redefined by the user.")
>  
> +;; Declare Lisp indentation rules for constructs that take 1 or 2
> +;; parameters before a body of RX forms.
> +(dolist (sym '( group-n submatch-n = >=))
> +  (put sym 'lisp-indent-function 1))

This is problematic: It changes the indentation behavior of the global
functions, depending on whether "rx.el" has been loaded.  Multiline
calls of `=' or `>=' might be rare, but multiline `rx' subexps are rare
as well.

Can we delay such changes until we have some kind of context sensitive
indentation algorithm for Elisp?

TIA,

Michael.

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Mattias Engdegård]

30 juni 2022 kl. 13.30 skrev Michael Heerdegen <michael_heerdegen@web.de>:

> This is problematic: It changes the indentation behavior of the global
> functions, depending on whether "rx.el" has been loaded.  Multiline
> calls of `=' or `>=' might be rare, but multiline `rx' subexps are rare
> as well.

Thank you, it was an overreach. Will revert.

I'm assuming that setting indentation properties on the other symbols, `group-n`, `submatch-n`, `**` and `repeat`, is less objectionable.

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Michael Heerdegen]

Mattias Engdegård <mattiase@acm.org> writes:

> I'm assuming that setting indentation properties on the other symbols,
> `group-n`, `submatch-n`, `**` and `repeat`, is less objectionable.

Not sure.  Still effects `let' bindings:

(let ((repeat x
          y
        z)))

Oh - and `repeat' is even a command.

Michael.

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Visuwesh]

[வியாழன் ஜூன் 30, 2022] Michael Heerdegen wrote:

> Mattias Engdegård <mattiase@acm.org> writes:
>
>> I'm assuming that setting indentation properties on the other symbols,
>> `group-n`, `submatch-n`, `**` and `repeat`, is less objectionable.
>
> Not sure.  Still effects `let' bindings:
>
> (let ((repeat x
>           y
>         z)))
>
> Oh - and `repeat' is even a command.
>

Not sure if this is of much significance here but 'repeat' also appears
in defcustom :type.

> Michael.

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Mattias Engdegård]

30 juni 2022 kl. 15.12 skrev Michael Heerdegen <michael_heerdegen@web.de>:

> Still effects `let' bindings:
> 
> (let ((repeat x
>          y
>        z)))

Someone's changed the syntax of let bindings without telling me!

> Oh - and `repeat' is even a command.

Hardly one likely to be called from Lisp though.
However it does see some use as a defcustom type so I've elided it as well.

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Stefan Monnier]

> I'm assuming that setting indentation properties on the other symbols,
> `group-n`, `submatch-n`, `**` and `repeat`, is less objectionable.

`group-n`, `submatch-n` sound specific enough, but the other two seem
too generic to my gut-feeling,


        Stefan

Re: master 3447e79f24 1/3: * lisp/emacs-lisp/rx.el: Indent some rx constructs correctly.

[Mattias Engdegård]

30 juni 2022 kl. 16.46 skrev Stefan Monnier <monnier@iro.umontreal.ca>:

> [`repeat` and `**`] seem too generic to my gut-feeling,

`**` doesn't seem excessively popular as function name so I'm keeping the rx indentation for now.

(But it would be a cool alternative name for `expt`. Which naturally gives us `***` for tetration, very useful!)