On Fri, Aug 26, 2016 at 4:53 AM, Christoph M. Becker <cmbecke...@gmx.de> wrote: > On 25.08.2016 at 18:37, Davey Shafik wrote: > >> On Thu, Aug 25, 2016 at 8:12 PM, Christoph M. Becker <cmbecke...@gmx.de> >> wrote: >> >>> Indeed, the RFC explicitly claims: >>> >>> | Behavior of internal functions is not going to be changed. >> >> This is correct for functions that had the correct behavior before >> (everything but array functions). >> >> I can remove that part of the change — but if we're going to change it, >> doing in 7.1 rather than > 7.1 would be best. > > AIUI, the actual implementation of the too_few_args RFC does what it > says, namely to not affect the behavior of internal function at all. > The issue you've raised is that we have introduced an inconsistency, > because some internal functions (e.g. array_diff) behave differently. > > If that is so, reverting the too_few_args RFC is out of question, > because not changing the behavior of internal functions was intentional, > and I assume that everybody who voted on the RFC was aware of that. > > The inconsistency between internal functions wrt. to too few/many > arguments in strict_types mode is not related to the too_few_args RFC at > all – we have this as of PHP 7. Fixing this inconsistency as soon as > possible might be desirable, but it seems to me that is not even > possible in the general case, as not all relevant code is under our > control (think of PECL and even "private" extensions). > > So, even we "fix" the behavior of all bundled internal functions, we > still can't claim that *all* functions throw an exception when called > with too few or too many arguments in strict_types mode (besides that we > easily might miss a few cases). If the docs already make this claim, > the docs should be fixed. > > Finally, I wonder why array_diff(), for instance, even has an explicit > check for ZEND_NUM_ARGS() and for Z_TYPE() != IS_ARRAY instead of > properly invoking zend_parse_parameters() with "aa+" instead of "+" in > the first place? Maybe I'm missing something, but otherwise I would > suggest to fix that altogether instead of piecemeal, even if that has to > wait until 8.0. > > -- > Christoph M. Becker > > -- > PHP Internals - PHP Runtime Development Mailing List > To unsubscribe, visit: http://www.php.net/unsub.php
Well, its behavior does differ with a single array: it will preserve keys in this case. If more than one array is passed it will numerically index the keys starting at zero. This is the only reason I can think of. -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
