[TYPO3-50-general] TypoScript 2.0 EBNF

Robert Lemke robert at typo3.org
Tue Jun 12 10:46:28 CEST 2007


Hi Elmar,

Am 12.06.2007 um 10:27 schrieb Elmar Hinz:

> Hmm, lexers also use regular expressions to tokenize. The API of  
> the lexer
> is quite simple: "Give me the next token." It's the token source  
> for the
> compiler. There is no need to reinvent regular expressions as an own
> finite state machine, even if you would use a fullblown lexer/compiler
> architecture.

Yes, I know. What I didn't emphasize enough was that my goal was to skip
tokenizing at all.

> Yes and no. It works. But it's a one step architecture. Separtion into
> different steps (preprocessor, lexer, compiler, ...) has the main  
> target to
> keep each step as simple as possible. And simple steps are easy to
> improve or extend or to put additional steps in between.

Well, my motivation to make it a one-step parser was simplicity - I  
saw that
it can be done for TypoScript because it didn't result in a complex  
code.

> Old TYPO3 IMHO died because of to less separation of functionality.  
> Every
> improvemt or extension of the giant, complex objects and functions  
> led to
> conflicts. It was to difficult to modernize it on time.

Absolutely. Honestly, if you have the impression that my current  
parser implemenation
is too complex or couldn't be changed easily, I have to rethink my  
approach. I'm
open for your suggestions.

>> Therefore most of the traditional conditions like browser agent, ip
>> address etc.
>> wouldn't work.
>
> At least if you don't want to parse it for each call. :-)

Yes. Of course we want to cache the TS object tree, that's also why I  
think parsing
speed is not that important yet.

>> Most of the other conditions (time, browser agent etc.) will have to
>> be implemented
>> with .if conditions. But at least we have simplified the syntax of  
>> that.
>
> So parse conditions could be implemented as a preprocessor.

Yes.

> A chance to separate them from TS and to keep TS clean.

I guess that can be done, yes. But then the same is true for  
variables - which are
one argument tokenizing.

robert
-- 
http://typo3.org/gimmefive




More information about the TYPO3-project-5_0-general mailing list