php.net |  support |  documentation |  report a bug |  advanced search |  search howto |  statistics |  random bug |  login
Request #64320 Proposal: deprecate the "leading 0 means octal" behaviour
Submitted: 2013-02-28 13:41 UTC Modified: 2013-02-28 18:08 UTC
From: php at richardneill dot org Assigned:
Status: Closed Package: Math related
PHP Version: 5.4.12 OS: All
Private report: No CVE-ID: None
 [2013-02-28 13:41 UTC] php at richardneill dot org
Description:
------------
PHP assumes that a string with a leading zero should be interpreted as octal.

In my experience, this is almost never useful, desirable, or intentional, however, it is a frequent trap for the beginner (or more experienced programmer), especially when parsing user-submitted data.

The only reason for this behaviour seems to be historical, and compatibility with strtol(). When non-programmer humans write numbers with leading zeros, they don't usually mean base 8.

My proposal is:

1. Introduce a new string format, 0o####  (letter o), to explicitly mean "this is an octal number". This is similar to the recent introduction of 0b#### for binary numbers. 

[This part should be simple to do, and would also make the intentional use of octal notation explicit, increasing code-readability]


2. Add an option to raise E_NOTICE any time a number is implicitly interpreted as octal (for example  "$x = 0100").


3. In a few years time, (PHP 7?), it may finally be possible to change the default behaviour.

Test script:
---------------
Here's an illustration of a naive program that has data-dependent bugs that are really hard to track down.

$mass_kg = "0.314"      //user-submitted data, known to begin "0."
$mass_g  = substr ($mass_kg, 2);

Yes, this is a silly thing to write. But it's a nasty trap for the unwary.
The above example does what is expected, and might pass many tests. But if the user subsequently enters $mass_kg = "0.031", this will turn into 25 grams, not 31 grams. As a result, we have created a very subtle and hard to find bug.




Patches

Add a Patch

Pull Requests

Add a Pull Request

History

AllCommentsChangesGit/SVN commitsRelated reports
 [2013-02-28 17:14 UTC] nikic@php.net
-Status: Open +Status: Feedback
 [2013-02-28 17:14 UTC] nikic@php.net
Could you maybe provide a testcase for the behavior you are referring to? I tried out the following and "010" wasn't interpreted as octal in any case:

var_dump((int) "010", intval("010"), "010" == 8);
// Outputs: int(10) int(10) bool(false)

010 is only treated as an octal if you write it out plainly (in the source code), like $foo = 010;
 [2013-02-28 18:08 UTC] php at richardneill dot org
You know, that bug has annoyed me for ages, since I got badly bitten by it about 7 years ago. In the meantime, I think it's been fixed! At any rate, I now agree with you that the current behaviour is correct, and that:

PHP now treats strings with leading zeros (eg "0123") as decimal (123), rather than octal, even when automatically casting them. 

The documentation (in the sections: integers, casting) could perhaps be more explicit, given that:

$a = 0x123;
$b = "0x123" + 0;     // $a and $b are equal
$c = 0123;
$d = "0123" + 0;      // $c and $d are not equal.

Sorry for the confusion - my bad.
 [2013-02-28 18:08 UTC] php at richardneill dot org
-Status: Feedback +Status: Closed
 
PHP Copyright © 2001-2024 The PHP Group
All rights reserved.
Last updated: Thu Mar 28 10:01:26 2024 UTC