[Typo3-dev] Validating TypoScript - Schema vs. DTD

[dan frost]

Hi everybody!

Here goes... I've spent ages looking at 
TypoScript, its implementation and architecture 
and alternatives. The following is my opinion... 
and (PLEASE READ THIS) i am not attacking anyone 
in-particular... I am making points which i think 
are assumed and important to realise about the 
whole TS issue.

1. I suggest that TypoScript as-is should not have 
more validation effort invested in it. A more 
object based model would justify such a thing

2. Remeber: TypoScript is (or was) meant to be 
configuation language - therefore a simple DTD 
would make sense for it

2a. But: the DTD (or whatever) should be 
extendable - i must be able to add 
MY_NEW_MENU_TYPE in the future and .my_silly_function

3. I have often suggested as twin approach to TS / 
XML - write in TS, but store (and describe) in 
XML. Therefore, you win the detail of XML (+ 
validation) and the easy of typing of TS.

4. In my opinion - and I *know* this is argued 
against - there is more worth in spending time of 
defining what TS should be and rebuilding it than 
in trying to build tools around it as is. (Please 
don't argue about my point of view - argue about 
the detail...)

5. Debugging TS *is* complicated and slow - anyone 
who says otherwise hasn't used profession 
development tools with break-points, full logging, 
object-tracking etc... or they have and can't 
imagine them in TS! ... my point: TS should be the 
fastest thing to debug: it must be tool, not a burden.

6. The whole "for the beginner" argument, where 
the beginner has to spend ages learning TS does 
not hold water: software is only as useful as the 
time it saves. If it takes 6 months to learn TS, 
and over 1000 developers are using it that meants 
that 500 developer-years have been spent learning 
it. IF this can be reduced, it should - the 
difficulty of a language is NOT a measure of its 
usefulness. Often, the opposite.

7. I think TS is a great model for the future of 
web development - small objects, small tools 
giving great power.

Now:
Do not argue. Do develop.

dan



Elmar Hinz wrote:
> Hello,
> 
> a month ago we had a discussion about the future of TypoScript. We came 
> to the point that it is important to validate it. Without validation we 
> get less error messages about wrong compositions. That slowes down 
> production. This slowdown is compensed by a very short form of notation 
> in comparism to XML. But we don't feel really lucky without the 
> possibility of validation. A lot of TYPO3 users would prefer XML.
> 
> Validation is an important point. But the discussion came to no real 
> result if validation of TypoScript could be possible. I guess the reason 
> for this is that only few of us really know how validation works.
> 
> I currently make some thoughts about converting TS to XML and to 
> validate that.
> 
> 
> page = PAGE
> page.10 = HTML
> page.10.value = Test
> page.10.if.value = 1
> page.10.if.equals = 2
> page.20 = COA
> page.20.10 = HTML
> page.20.10.value = Test
> 
> ...
> 
> <page type="PAGE">
>   <10 type="HTML">
>     <value>Test</value>
>     <if>
>       <value>2</value>
>       <equals>1</equals>
>     </if>
>   </10>
>   <20 type="COA">
>     <10 type="HTML">
>       <value>Test</value>
>     </10>
>   </20>
> </page>
> 
> I don't know if numbers are valid XML Elements. But one could write 
> <number10> instead of <10>. That isn't the point.
> 
> I see a problem with ambiguus elements like "value" or "number10". 
> Depending on their context they are of different Types. That means that 
> depending on their own parents different childs are allowed.
> 
> My knowladge about DTD tells my, that is not possible to define 
> different childs for an element depending on the parent context.
> 
> Does anybody know if schema do offer something like this?
> 
> Regards
> 
> Elmar
> 
> 
> 
> 
> 
> 
>