[TYPO3-dev] How to use the new autoloader feature?

[Xavier Perseguers]

Hi Ingo,

> On serialized arrays: Why serializing and deserializing on every page 
> request when you can have the native and human readable format? (saving 
> calls to deserialize())

If you think in term of parsing time, both the serialized array and the 
human-readable version need time to be "transfered" into memory because 
none of them may be copied 1-1 to some memory slot before being usable.

The thing is that when you have a human-readable array, you need to 
parse it, remove spaces, find out the length and place it will take into 
memory for each element, ... whereas a serialize array already has all 
those information at hand making the "unserialize" call much more a 
human-readable function that tells the parser/analyzer and then compiler 
"hey, don't bother. The work already is done, just enjoy!".

I did not make any benchmark and I guess the difference is hardly 
distinguishable but it would be quite easy to benchmark both versions 
with a few hundreds of elements and of course no cache-code library in use.

Regards

-- 
Xavier Perseguers
http://xavier.perseguers.ch/en

One contribution a day keeps the fork away