Add `\f` as a whitespace character in `trim`, `ltrim` and `rtrim`

[LamentXU123]

Description

Hi.
The trim function will remove whitespaces in the parameters. The default characters are \n\r\t\v\0, source.
I think maybe \f is missing here? In most of the cases in php (such as is_numeric, source) we recognized \f as a whitespace. Also in libc isspace.c we too include \f as whitespaces:

/*-
 * See the file LICENSE for redistribution information.
 *
 * Copyright (c) 2005,2007 Oracle.  All rights reserved.
 *
 * $Id: myisspace.c,v 1.4 2007/05/17 15:14:54 bostic Exp $
 */

 /*
  * myisspace --
  *
  * PUBLIC: #ifndef HAVE_ISSPACE
  * PUBLIC: int myisspace __P((int));
  * PUBLIC: #endif
  */
int
myisspace(c)
int c;
{
	return (c == '\t' || c == '\n' ||
		c == '\v' || c == '\f' || c == '\r' || c == ' ' ? 1 : 0);
}

Since the document indicate that trim would "Strip whitespace (or other characters) from the beginning and end of a string", I think \f may be an unexpected case here as it is usually treated as whitespaces. Thanks.

PHP Version

all version

Operating System

No response

[devnexen]

python, rust ... to mention a handful, trim whitspaces, form feed included. it sounds legit to expect the same here.

[devnexen]

btw since you understood well the problem, feel free to open a PR ;)

[Girgias]

While I agree this makes sense, you may still want to send a quick email to the Internals mailing list to make sure nobody complains, as it is technically a BC break.

[LamentXU123]

btw since you understood well the problem, feel free to open a PR ;)

Sure. The PR is submitted: #20788
btw I change this issue to three functions by adding ltrim and rtrim. I think they are same as trim in the issue.

While I agree this makes sense, you may still want to send a quick email to the Internals mailing list to make sure nobody complains, as it is technically a BC break.

It is sent XD.

[bukka]

By our policy, all BC breaks need to go through RFC.

[LamentXU123]

By our policy, all BC breaks need to go through RFC.

If strictly adhering to POSIX/C locale definitions (specifically isspace()), it functions as a whitespace character. I don't find RFC giving direct definitions to "whitespace". However, What I do find is the definition of whitespace in HTTP protocols in RFC 7230 / RFC 9110, which strictly defined "whitespace" in HTTP protocols as SP(\x20) and HTAB(\t). I don't think this is suitable in this case because we've already include \n\r\t\v\0 as whitespaces and it is not only for HTTP usage.
I don't think we've got specific RFC definitions of this case. However we do get POSIX/C locale definitions of whitepaces in isspaces() which contains \f. Also comparing to other languages \f is often treated as "whitespace".
such as in python:

[LamentXU123]

Another source of this is from rust's documentation about trim()

Returns a string slice with leading and trailing whitespace removed. 'Whitespace' is defined according to the terms of the Unicode Derived Core Property White_Space.

By looking at the definition of "Unicode Derived Core Property White_Space". Form Feed is in this definition:

Hence I think this feature make sense to me.

Also in trim() funtion of javascript, we too treat \f as "whitespaces". See: ECMA-262