[kairosci]

Fixes #19509.

[anun333]

Quoted below: written with Claude, an LLM, as CONTRIBUTING.md asks us to mark. The builds and test runs behind it are on my machine; the commands to repeat them are under the fold.

Tested on today's master (17cfa007d5d) with this applied, debug builds, display_errors=1, counting messages shown:

	master	this PR
the issue's loop, E_ALL: the warning / the deprecation after it	10 / 10	1 / 1
the same, E_ALL & ~E_DEPRECATED	10	1
ignore_repeated_errors=0 (control)	10	10
one message from two lines, ignore_repeated_source=0 / 1	6 / 1	6 / 1
error_get_last() after the loop	the deprecation	the deprecation
a user error handler, calls	20	20
Zend/tests and tests	6,448 passed	6,449 passed, 0 failed

One design note: this keeps a last message per error type, so a repeat is also suppressed when other displayed errors come in between (the deprecation above drops to one too). The issue's report shows only warnings, so its deprecations were filtered out by error_reporting; tracking just the last error that passed error_reporting, as suggested on the issue, should cover that case without widening what counts as repeated. Which of the two is wanted is a call for the reviewer.

How this was tested

Linux x86_64, gcc 13.3, debug builds of origin/master and of it with this PR's commit cherry-picked:

./buildconf --force
./configure --enable-debug --disable-cgi --disable-phpdbg --disable-fileinfo --enable-pcntl --enable-zend-test --enable-fpm --with-curl --enable-mbstring --disable-mbregex
make -j4

Each row: sapi/cli/php -n -d display_errors=1 -d html_errors=0 -d ignore_repeated_errors=1 -d error_reporting=… -r 'for ($i = 0; $i < 10; $i++) { trim($at); }', counting "Undefined variable" and "Deprecated" lines; the source rows use echo $a; on two lines, three times. Suites: TEST_PHP_EXECUTABLE=$PWD/sapi/cli/php sapi/cli/php run-tests.php -n -q -j4 Zend/tests tests.

[eyupcanakman]

I built this and its parent (e92d9c4) to compare. The quoted part was written with an LLM, per CONTRIBUTING.

It fixes the reported loop. Three other cases change, and no existing test catches them. Each case runs as its own script with -d ignore_repeated_errors=1.

function a() { trigger_error("A", E_USER_WARNING); }
function n() { trigger_error("N", E_USER_NOTICE); }

// 1. master shows A twice, this PR once
a(); error_clear_last(); a();

// 2. master shows all 6, this PR shows A and N once each
for ($i = 0; $i < 3; $i++) { a(); n(); }

// 3. master prints "A", this PR prints "D"
error_reporting(E_ALL & ~E_USER_DEPRECATED);
a(); trigger_error("D", E_USER_DEPRECATED); a();
echo error_get_last()['message'];

The first two come from the per-type slots. error_clear_last() doesn't reset them, and today a visible error only counts as a repeat when it comes right after the same one.
In the third, the second a() is hidden, so error_get_last() still returns the deprecation.