Hello Andi, Monday, December 1, 2003, 6:49:09 PM, you wrote:
> Hey, > Let's try and summarize this discussion and try and see what solution we > can come up with. > The change which was made in convert_to_string() is quite problematic. It > touches a very fundamental engine function and changes the behavior of tons > of places. For example: > a) include $obj; will not error out but convert the object to a string and > then include it. > b) $obj[2] = 3; (where $obj implements a __toString function) isn't quite > defined anymore. Should this work? > c) $arr[$obj] would suddenly work according the __toString(). > Not to say that all sorts of bugs would probably be introduced because all > places which call convert_to_string() would have to be checked. > The second patch which automatically calls __toString() on parameters isn't > quite as bad but I am really not sure this is what we want. Do we want PHP > scripts to become much less readable and have people save their strings in > objects and pass those around? How will passing such objects by reference > work with extension functions which expect their string parameter to be by > reference? > As you can see there are LOTS of problems this would introduce (and most > likely more than I could think of). > In the past it was agreed that > __toString() would be called explicitly but we would let "print" to be > overloaded. > Right now, I think making people call it explicitly is the best way to go > and the way Java chose too (not that I thnk we necessarily need to go the > same way but there were some smart ppl who thought of these issues before us). > I do think that SimpleXML has to be tailored for because I truly think that > it's the best addition to PHP 5 including the new OOP features. > So the question really boils down to what exactly the SimpleXML extension > requires? Can't it return a string when it's in read mode? > Let's try and find the right balance between ease-of-use for extensions > such as SimpleXML and verbosity and a PHP which is not error-prone and as > consistent as possible. > Marcus, reverting the patch was nothing personal. > You can see I have technical issues with it. I think you have been doing > some great patches and you know it :) The summary of my productive thoughts on this: Hmmm, the minimum needed is to allow: whatever((string)$obj) then. The solution to this would be another object handler, say make_printable() and move the __toString() Call from the cast_object handler into this new one AND partly reintroduce my patch and voila all are happy? Maybe this is a really cool idea because it allows all kinds of 'cool' objects to be written in C, it allows all kinds of language bindings but without the HIGH __toString() WTF factor (minimal compareable to C++ autoconversion problems for example). -- Best regards, Marcus mailto:[EMAIL PROTECTED] -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
