No one? On Mon, Feb 2, 2015 at 8:57 PM, Juan Basso <jrba...@gmail.com> wrote:
> Hi, > > I was testing few things today and I released one bug in the serialize > function. The serialize function returns a corrupted value when it is > serializing an object that contains a __sleep magic method and this method > return some key that are not string. Ie: > > http://viper-7.com/EK16EV > > If the __sleep returns a string that is not a property it is fine, it will > give the notice and the generated response is a valid serialized data (in > another words you can use unserialize to retrieve the object back). Ie, > http://viper-7.com/H9ooff > > I can try to make a patch to solve it, but before that I would like how > the behavior should be. Some options: > 1) Give the notice saying the field doesn't exist and do not include on > the serialized response > 2) Give the notice saying the field doesn't exist and convert the value to > string (ie, on my example the int(1) would be transformed to string(1)) > 3) Give a warning and return false once the input from __sleep is invalid > 4) None of the above alternative > 5) Don't change it > > > Thanks, > Juan Basso >
