Re: Re: important bug, please read

Stanislav Malyshev wrote:

AL>> don´t do that!!! that´s what we (including you) are fighting against, AL>> "reference to..." is evil :) AL>> why not simlply AL>> AL>> array(3) {

AL>>     [0]=> int{&1}(1)
AL>>     [1]=> int(2)
AL>>     [3]=> int{&1} }
AL>> AL>> only variables with ref_count>1 will be attached on {} index thing and 

No good. Refcount for variable can have arbitrary values in given point. Think copy-on-write. Lots of it.

umh, sure - I actually haven´t seen the patch, but you should be able to buffer your output like this: (replace [~]n(i) with an unique string for every zval ((I assumed you do not use XML internally :) ))) array(3) { [0]=> int{[~]1}(1) [1]=> int{[~]2}(2) [3]=> int{[~]1} } then count these [~]n artefects to calculate your own "ref_count", if there´s only a single artefect [~]n(i) delete the whole brackets {[~]n(i)} if there are more replace [~]n(i) with an incremented number and your output as follows: array(3) { [0]=> int{&1}(1) [1]=> int(2) [3]=> int{&1} } well, I think there are many better ways to realize, just a thought regards