|
php.net | support | documentation | report a bug | advanced search | search howto | statistics | random bug | login |
[2012-01-10 02:26 UTC] MarkAndrewSlade at gmail dot com
Description:
------------
Unlike echo, the print construct is allowed inside logic clauses. If it is
reached (not short-circuited), it will cause the rest of that clause to be true.
Tested with 5.3.8 and PHP 5.3.9RC5-dev, both with default configure.
Test script:
---------------
<?php
echo (false && print('')) ? "Fail\n" : "Pass\n";
echo (print('') && false) ? "Fail\n" : "Pass\n";
echo (print('') && false && false) ? "Fail\n" : "Pass\n";
echo ((print('') && false) && true) ? "Fail\n" : "Pass\n";
echo ((print('') && false) && false) ? "Fail\n" : "Pass\n";
?>
Expected result:
----------------
The word "Pass" five times.
Actual result:
--------------
The middle three fail. The first and last pass, and are included to demonstrate
the limits of the bug.
PatchesPull RequestsHistoryAllCommentsChangesGit/SVN commits
|
|||||||||||||||||||||||||||
Copyright © 2001-2025 The PHP GroupAll rights reserved. |
Last updated: Thu Oct 30 03:00:01 2025 UTC |
Invalid interpretation of the code caused by misleading parenthenses. 1. false && print ('') -> false && (print ('')) -> false && 1 -> false 2. print('') && false -> (print (('') && false)) -> (print false) -> 1 -> true 3. print('') && false && false -> (print ((('') && false) && false)) -> (print false) -> 1 -> true 4. (print('') && false) && true -> ( print (('') && false) ) && true -> (print false) && true -> 1 && true -> true 5. (print('') && false) && false -> (print (('') && false) && false -> (print false) && false -> 1 && false -> false I believe that documentation for all language constructs (`echo`, `include`, `print`, ...) should explicitly discourage use of parenthenses around arguments. They're very misleading.Oh, I see what happened. I agree with your recommendation. The documentation says they are "not required" (implying optional), and they are used in the actual example. I normally use echo, and without parens, so I just kinda randomly came across this. I'm not sure if you were saying the parser's interpretation was invalid or mine, but in case the latter and this is considered correct parsing, the documentation should be updated to reflect that it's "print <string>", without parens. Developers can deduce for themselves that "print ('foo')" is allowed, but the semantics will be clearer.