cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nicola Ken Barozzi <nicola...@apache.org>
Subject Re: [RT] the quest for the perfect template language
Date Fri, 04 Apr 2003 15:28:33 GMT


Stefano Mazzocchi wrote, On 04/04/2003 16.56:
> Nicola Ken Barozzi wrote:
...
>> So it's just style, right?
> 
> style, mindset, syntax sugar, call it as you like. but a perfetly good 
> concept (RDF, for example) can be ruined by lack of style (its xml 
> syntax), just like a nice engine can be totally hidden inside an 
> ugly-looking car.

Personally I don't think that RDF's problem is syntax... it's the lack 
of something that needs RDF to run and I need that makes me understand I 
think.

> gosh, you are the first italian I ever heard that doesn't care about 
> style :-)

;-P   I'm in my personal phase of KISS... maybe call it DFWSS (don't fix 
working stuff, stupid). I don't want to spent too much time reinventing 
stuff for marginal gains.

> Guess why i bought a mac :-)

I know, you were salivating already a couple of years ago ;-)

>>> [note, since stylesheet are compiled in memory anyway, the extra 
>>> transformation doens't add any performance problems at runtime]
>>  
>> I'd like to see that before believing. XSPs were compiled, hence the 
>> fastest... but oddly enough the interpreted sitemap is faster still.
> 
> There is nobody who is more blind that somebody who doesn't want to see.
> 
> Saint Thomas, look: if the syntax adaptation stage is done *before* the 
> execution, what difference can it possibly make at runtime?

It all depends to *how* you do the adaption. The fact that C can 
optimize code when it compiles doesn't say anything about how well it's 
optimized. And you know how the cimpiled sitemap started suffering badly 
and becoming black art when things got more and more complicated.

So, I still have to see it. ;-)  My KISS tells me that 'more' is usually 
not a simplification.

>>>> XSLT? I like it. For simple transformations IMHO it really rocks. 
>>>> With a relative small number of tags and some xpath it does almost 
>>>> all that is needed.
>>>
>>>
>>>
>>> The cost of writing a stylesheet is exponential with time and with 
>>> people involved. I want to solve this.
>>
>> And that's because of pointy brackets? The same holds true for Java or 
>> any language.
>>
>> XML has to be valid, and I like to be able to validate my XSLT 
>> stylesheet. 
> 
> The parser that is responsible for transforming the non-xml syntax to 
> the xml one can be *MUCH* more validating than any xml schema language 
> and, for sure, much more precise in giving you error messages.

Ah, ok, so we need another validator, another set of tools... I just 
need well-formdness or DTDs 99% of the time, not some fancy schema 
stuff. And there is so much XML stuff out there that the switch doesn't 
seem compelling.

>> I tried Velocity, but I kept outputting erroneously non-wellformed 
>> XML, and after a while I got fed up with it totally.
> 
> Velocity is text-based, my approach is sax based. Huge difference.

Jelly is SAX based. We could call it SAX Velocity.

> velocity has a nice syntax, but it's text-baseness is dead poor in a 
> sax-intensive environment like ours. besides the requirement of an extra 
> processing stage which blows the performance.
> 
> the idea to use a non-xml syntax for XSLT would allow the best of both 
> worlds:
> 
>   - friendly syntax
>   - structured results
> 
> I really don't understand why some of you are so emotionally attached to 
> something like
> 
>  <xsl:if test="count(blah) &gt; 3">

And I'm surprised why this thing makes you desire such a massive 
endeavour so much. I don't see value for the "money".

> but even more I'm surprised to see 'conservationism' on this list.
> 
> Are you guys getting old or shy or what? ;-)

We are getting a bit tired I guess of running after changes that give us 
more work than benefits.

-- 
Nicola Ken Barozzi                   nicolaken@apache.org
             - verba volant, scripta manent -
    (discussions get forgotten, just code remains)
---------------------------------------------------------------------


Mime
View raw message