Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:129000
Precedence: list
MIME-Version: 1.0
Date: Wed, 29 Oct 2025 09:56:07 -0700
Message-ID: <CAE1541uqb8qZa967UJNBXO2OkmM9VDbCJhZ+ruy=EyBP8+FfCQ@mail.gmail.com>
Subject: [PHP-DEV] Review request: Enabling `transitive_compare_mode` to fix SORT_REGULAR
To: PHP internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="000000000000d7656006424f0224"
From: jmarble@intuitivetechnology.com (Jason Marble)

--000000000000d7656006424f0224
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hello PHP Internals!

I believe I've found an acceptable solution to the issue that's plagued
SORT_REGULAR since the dawn of time.

Fundamentally, all sorting algorithms require transitive comparisons (if A
=E2=89=A4 B and B =E2=89=A4 C, then A =E2=89=A4 C). However, PHP's SORT_REG=
ULAR comparison function
violates transitivity when mixing numeric strings, non-numeric strings, and
numbers, leading to unpredictable and non-deterministic sort results.

**The Solution:**
Add a `transitive_compare_mode` flag to executor_globals (TLS) that signals
zendi_smart_strcmp() to enforce consistent ordering during SORT_REGULAR
operations:
- Numeric strings are consistently ordered relative to non-numeric strings
- Eliminates circular comparisons
- Maintains PHP 8+ semantics (numeric-types < numeric-strings < non-numeric=
)

**Implementation:**
- PR: https://github.com/php/php-src/pull/20315
- Uses save/restore pattern for reentrancy safety
- New test coverage included

**Historical Context:**
Raghubansh Kumar documented this issue in 2007, creating tests with the
note "(OK to fail as result is unpredectable)". Nikita Popov's 2019 RFC
improved string-to-number comparison semantics but didn't eliminate the
transitivity violation. This fix completes that work.

**Test Results:**
Four variation11 tests currently fail as expected, but their outputs are
now deterministic and more "sane" than both the 2007 and 2019 versions imo:
https://gist.github.com/jmarble/957a096cb2bf25b577de47449305723f

**ABI Considerations:**
This adds a field to _zend_executor_globals, which is technically an ABI
break appropriate for PHP 8.6. However, practical impact is minimal, I
think, since most extensions access executor_globals through the EG() macro
(which abstracts the struct layout),
not via direct struct manipulation.

I'd appreciate review and feedback on this approach. This is a
long-standing correctness issue.

Thank you for your time and consideration!

- Jason

--000000000000d7656006424f0224
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Hello PHP Internals!</div><div><br></div><div>I belie=
ve I&#39;ve found an acceptable solution to the issue that&#39;s plagued SO=
RT_REGULAR since the dawn of time.</div><div><br></div><div>Fundamentally, =
all sorting algorithms require transitive comparisons (if A =E2=89=A4 B and=
 B =E2=89=A4 C, then A =E2=89=A4 C). However, PHP&#39;s SORT_REGULAR compar=
ison function violates transitivity when mixing numeric strings, non-numeri=
c strings, and numbers, leading to unpredictable and non-deterministic sort=
 results.</div><div><br></div><div>**The Solution:**<br>Add a `transitive_c=
ompare_mode` flag to executor_globals (TLS) that signals zendi_smart_strcmp=
() to enforce consistent ordering during SORT_REGULAR operations:<br>- Nume=
ric strings are consistently ordered relative to non-numeric strings<br>- E=
liminates circular comparisons<br>- Maintains PHP 8+ semantics (numeric-typ=
es &lt; numeric-strings &lt; non-numeric)</div><div><br></div><div>**Implem=
entation:**<br>- PR: <a href=3D"https://github.com/php/php-src/pull/20315">=
https://github.com/php/php-src/pull/20315</a><br>- Uses save/restore patter=
n for reentrancy safety<br>- New test coverage included</div><div><br></div=
><div>**Historical Context:**<br>Raghubansh Kumar documented this issue in =
2007, creating tests with the note &quot;(OK to fail as result is unpredect=
able)&quot;. Nikita Popov&#39;s 2019 RFC improved string-to-number comparis=
on semantics but didn&#39;t eliminate the transitivity violation. This fix =
completes that work.</div><div><br></div><div>**Test Results:**<br>Four var=
iation11 tests currently fail as expected, but their outputs are now determ=
inistic and more &quot;sane&quot; than both the 2007 and 2019 versions imo:=
<br><a href=3D"https://gist.github.com/jmarble/957a096cb2bf25b577de47449305=
723f">https://gist.github.com/jmarble/957a096cb2bf25b577de47449305723f</a><=
/div><div><br></div><div>**ABI Considerations:**<br>This adds a field to _z=
end_executor_globals, which is technically an ABI break appropriate for PHP=
 8.6. However, practical impact is minimal, I think, since most extensions =
access executor_globals through the EG() macro (which abstracts the struct =
layout),<br>not via direct struct manipulation.</div><div><br></div><div>I&=
#39;d appreciate review and feedback on this approach. This is a long-stand=
ing correctness issue.<br><br>Thank you for your time and consideration!<br=
></div><div><br></div><div>- Jason</div><div dir=3D"ltr" class=3D"gmail_sig=
nature" data-smartmail=3D"gmail_signature"><div dir=3D"ltr"><div></div></di=
v></div></div>

--000000000000d7656006424f0224--