Bug #79698 [Opn->Csd]: timelib mishandles future timestamps (triggered by 'zic -b slim')

From: Date: Tue, 06 Apr 2021 19:56:27 +0000
Subject: Bug #79698 [Opn->Csd]: timelib mishandles future timestamps (triggered by 'zic -b slim')
References: 1  Groups: php.bugs 
Request: Send a blank email to php-bugs+get-233266@lists.php.net to get a copy of this message
Edit report at https://bugs.php.net/bug.php?id=79698&edit=1 ID: 79698 Updated by: derick@php.net Reported by: eggert at cs dot ucla dot edu Summary: timelib mishandles future timestamps (triggered by 'zic -b slim') -Status: Open +Status: Closed Type: Bug Package: Date/time related Operating System: any PHP Version: 7.4.7 -Assigned To: +Assigned To: derick Block user comment: N Private report: N New Comment: The fix for this bug has been committed. If you are still experiencing this bug, try to check out latest source from https://github.com/php/php-src and re-test. Thank you for the report, and for helping us make PHP better. Fixed for PHP 8.1. Previous Comments: ------------------------------------------------------------------------ [2020-06-13 20:03:03] nikic@php.net Can you please report this issue upstream at https://github.com/derickr/timelib/issues? ------------------------------------------------------------------------ [2020-06-13 20:00:27] eggert at cs dot ucla dot edu Description: ------------ Since tzdb 2019b, the 'zic' command has had a '-b slim' option that generates smaller-but-equivalent TZif files, intended to save memory/cache/whatever. These files conform to the TZif format specified in Internet RFC 8536 and are supported by tzcode, glibc, etc. and '-b slim' is intended to be the default in future tzdb releases. However, some current PHP timelib test cases fail with slim TZif files. This mishandling seems to occur because timelib gets confused about timestamps after the last explicit transition in a TZif file. For example, 'zic -b slim' generates a TZif file Europe/Amsterdam where the last explicit transition is on 1996-03-31 and transitions thereafter are deduced from the "CET-1CEST,M3.5.0,M10.5.0/3" string embedded at the end of the TZif file. The test case decoding 2008-03-30 01:00:00 fails, I expect because this timestamp is after the last explicit transition. If my guess is right, timelib also mishandles timestamps past the last explicit transition even in traditional "fat" TZif files. For example, since the last transition in a traditional Europe/Amsterdam file is 2037-10-25 I would expect timelib to mishandle some timestamps after that. As the '-b slim' option becomes more popular this problem will affect PHP even for past and current timestamps, so the bug needs to be fixed sooner than 2038. Test script: --------------- #!/bin/sh # This assumes a typical GNU/Linux host. wget https://data.iana.org/time-zones/releases/tzdb-2020a.tar.lz tar xf tzdb-2020a.tar.lz cd tzdb-2020a make make ZFLAGS='-b slim' posix_only # This needs root privileges! cd .. git clone https://github.com/derickr/timelib.git cd timelib make test Expected result: ---------------- I expect all tests in 'make test' to succeed. Actual result: -------------- Most tests succeed, but I see the following failures: transition.render FAIL | 2008-03-30 02:00:00 CEST Europe/Amsterdam | 1206835200 (Europe/Amsterdam) EXP = 2008-03-30 01:00:00 CET Europe/Amsterdam FAIL | 2008-03-30 02:59:59 CEST Europe/Amsterdam | 1206838799 (Europe/Amsterdam) EXP = 2008-03-30 01:59:59 CET Europe/Amsterdam ... render.render ... FAIL | 2005-10-30 03:00:00 CEST Europe/Oslo | 1130634000 (Europe/Oslo) EXP = 2005-10-30 02:00:00 CET Europe/Oslo ... FAIL | 2005-10-30 03:00:00 CEST Europe/Amsterdam | 1130634000 (Europe/Amsterdam) EXP = 2005-10-30 02:00:00 CET Europe/Amsterdam FAIL | 2005-10-30 04:00:00 CEST Europe/Amsterdam | 1130637600 (Europe/Amsterdam) EXP = 2005-10-30 03:00:00 CET Europe/Amsterdam FAIL | 2005-10-30 05:00:00 CEST Europe/Amsterdam | 1130641200 (Europe/Amsterdam) EXP = 2005-10-30 04:00:00 CET Europe/Amsterdam ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=79698&edit=1

« previous php.bugs (#233266) next »