[TYPO3-core] good question

[Kasper Skårhøj]

Yes, this strange construction was done in order to resolve any variable 
references in the process of copying TypoScript chunks. Basically the problem 
was that if someone did like:

page.10 = TEXT
page.20 < page.10
page.10 >

.... then internally page.20 would be a reference to page.10 and therefore 
both page.10 and page.20 would be deleted when page.10 was unset. Or another 
porblem: 

page.10 = TEXT
page.20 < page.10
page.10 = HTML

... then page.10 would also change page.20 because it was a reference.

BUT when I serialized/unserialized this would go away. I introduced this from 
PHP3 to PHP4.

My problem is this: Above I have explained that the problem was _something_ 
with references but I remember that it was only in some rather exotic and 
hard-to-find case that it failed. This means; If we change this we might not 
encounter problems before someone reports strange errors we don't believe he 
actually has - and that will create lots of confusion. To make a change we 
MUST create some UNIT test which tests all and can validate that maybe today 
PHP doesn't need this. Until then I'm for keeping it although it looks 
obscure...

- kasper



On Saturday 26 November 2005 16:42, Michael Stucki wrote:
> Hi Bernhard,
>
> > May somebody tell me what is the sense of the
> > "unserialize(serialize(...))" in this line:
> >
> >                     $this->setVal($objStrName
>
> $setup,unserialize(serialize($res)),1);
>
> > (line 289 of class.t3lib_tsparser.php)
> >
> > It seems as it does nothing except wasting cpu cycles :)
>
> I agree. I just checked: This has been there since CVS rev 1.1 so I suppose
> it was written by Kasper. Maybe he can explain?
>
> - michael

-- 
- kasper

-----------------
"A 'please' would be nice", John Travolta, Pulp Fiction