Problem with jewishtojd For Years >= 6000 for Php8.3+

[oleibman]

Description

The following code:

<?php
for ($yh = 5995; $yh < 6005; ++$yh) {
    $rh = jewishtojd(1, 1, $yh);
    echo "yh=$yh rh=$rh\n";
}

Resulted in this output for Php8.3, 8.4, 8.5:

yh=5995 rh=2537279
yh=5996 rh=2537633
yh=5997 rh=2538016
yh=5998 rh=2538371
yh=5999 rh=2538725
yh=6000 rh=0
yh=6001 rh=0
yh=6002 rh=0
yh=6003 rh=0
yh=6004 rh=0

But I expected this output instead (as it was for Php8.2, 8.1, 8.0, and 7.4):

yh=5995 rh=2537279
yh=5996 rh=2537633
yh=5997 rh=2538016
yh=5998 rh=2538371
yh=5999 rh=2538725
yh=6000 rh=2539110
yh=6001 rh=2539463
yh=6002 rh=2539818
yh=6003 rh=2540202
yh=6004 rh=2540557

PHP Version

PHP 8.5.0 (cli) (built: Nov 18 2025 08:17:59) (ZTS Visual C++ 2022 x64)
Copyright (c) The PHP Group
Zend Engine v4.5.0, Copyright (c) Zend Technologies
    with Zend OPcache v8.5.0, Copyright (c), by Zend Technologies

Operating System

Windows 11

[oleibman]

FWIW, Php8.3.6 on WSL (Windows subsystem for Linux) works properly. However, testing on some actual Linux machines indicate that they have the same problem as Windows.

[iluuu1994]

This seems to be on purpose. #18849 I can't argue for the rationale. @devnexen Can you clarify whether the limit can be raised?

[oleibman]

I'm beyond speechless! But I'll try anyhow.

You aren't supposed to introduce a breaking change except in a major release. This wasn't introduced in a major release. It wasn't even introduced in a minor release. It was introduced in a point release for 8.3.

The change isn't documented anywhere (no mention of any limits in the current documentation for jewishtojd).

I'll get to arguing about the appropriateness of your choice in a minute. But, before I get to that, let's assume for the sake of argument that setting a limit is a reasonable action. What should you be doing if the limit is exceeded - giving a wrong answer, or doing something sensible like throwing an exception? You have made the insane choice of being okay with giving a wrong answer!

The choice of limit is completely capricious. No serious thought seems to have been given to it at all. Is "I do not think there is a need to go too far into the future" really a justification? Did you consult with anybody? Is the "Messianic age" really something that deserves consideration in resolving this question? Is there no logical choice between 0 and infinity other than 200 years? (For the record, I'm glad to suggest one - the Jewish calendar repeats every 689,742 years, a value which, even when multiplied by days in the year, is not in any danger of overflow even on a 32-bit system.)

No need to go too far? Is there a similar throttle placed on Gregorian calendar? Julian calendar? Are they not subject to the same problem that was reported for the Jewish calendar? If I report that as a bug, will you break those in the same manner?

There are lots of tables that go far into the future. I can find tables of eclipses for many millennia to come. Should these be cut off too, because "you and I won't be here to see it anyway"?

This change is utterly broken. Please unbreak it.

[iluuu1994]

@oleibman Valid criticism. This is a somewhat older bug fix and we've since already discussed avoiding fixes for fuzzing issues in patches, assuming they are unlikely to have a real-world impact. Hopefully this reinforces that sentiment.