Bob, I see you already committed this into PHP-5.6.
I would agree that new behavior is more consistent and may be committed into master, but I'm still afraid that it may bring as problems because of last minute changes and lack of tests coverage. Thanks. Dmitry. On Sun, Jul 27, 2014 at 3:02 PM, Bob Weinand <bobw...@hotmail.com> wrote: > Am 27.7.2014 um 10:55 schrieb Stas Malyshev <smalys...@sugarcrm.com>: > > Hi! > > Yes, I agree that this is not correct behavior - and I don't really > understand why it was introduced and why it isn't trivial to fix. > PHP-5.5 had a check for this case in place > (http://lxr.php.net/xref/PHP_5_5/Zend/zend_compile.c#7071) and phpng > contains an AST-compatible variant of the array check > (http://lxr.php.net/xref/phpng/Zend/zend_compile.c#7776). Shouldn't > copying the condition from phpng into PHP-5.6 resolve this issue? > > > I agree it should be easy to fix it this way, but I'd like for Bob to > provide a bit more input here as to best way to resolve it. I'm not sure > why usage of arrays in runtime is disallowed now in 5.6 code, so I'm not > sure if we should enable it or remove it. > > If we don't find another way soon, I guess porting one from phpng is > what we'll have to do. > -- > Stanislav Malyshev, Software Architect > SugarCRM: http://www.sugarcrm.com/ > > > The AST compatible fix in phpng is just for top-level arrays, but not for > something like "constant ? [1] : [2]". > > I think we should just enable it, that would lower the level of confusion. > > I totally agree that current status isn't optimal. > > Bob >
