Re: [xml-dev] SGML complexity

[Rick Marshall <rjm@zenucom.com>]

ok, can't resist any longer.

I use xslt a lot. Because it is declarative it is incredibly easy to to 
do very complex things. (unibase is declarative too so at this stage i 
should admit to decades of experience with declarative programming).

Nevertheless the transformations that can be done with xslt are rich and 
at times beautiful.

eg I typically use xslt to write tcl scripts and postscript.

?

I have an xsl-fo type vocabulary for generating documents and use xlst 
to produce the postscript needed to draw the documents. (interestingly, 
at this stage pdf does not seem powerful enough).

Our edi systems work by using xslt to transform partner documents 
(written in a perverted xml vocabulary based on ansi vics850 order) to 
tcl which then populates our databases.

I guess that means you could just as easily use it to write sql (if you 
must).

Why xslt? well for starters the declaration of a template for the nodes 
of interest makes it easy to maintain a program that of itself mirrors 
the structure of the source documents. Debugging and extension are very 
easy. Yes you could do the same with sax, but my personal preference is 
declarative rather than procedural and I have expereinced elsewhere that 
it's easier to deal with very complex structures using declarative 
techniques than procedural techniques.

xslt rocks

(just thought i'd help tick up the counters ;) )

Rick

juanrgonzaleza@canonicalscience.com wrote:
> Alexander Johannesen said:
>   
>> On 9/7/06, juanrgonzaleza@canonicalscience.com wrote ;
>>     
>>> But sincerely do you think that larger number of articles and books
>>> devoted to PHP mean that XSLT is easier?
>>>       
>> Oh yes, by a large degree. XSLT as a language is substantionally
>> smaller, smarter, and more dedicated.
>>     
>
> Other strongly disagree with you. Any statistics at hand?
>
>   
>> To do certain things, there's only
>> a handful of ways. In PHP, there's infinite possibilities, by virtue of
>> being a full programming language.
>>     
>
> You see XSLT more focused to 'machine'. Whereas PHP opens the doors of
> creativity and this appears to be a serious trouble for you.
>
>   
>> XSLT is for
>> transformation, not programming, and hence don't need all that
>> complexity.
>>     
>
> XSLT can be considered a programming language in all right:
>
> <blockquote>
> As a programming language, XSLT has many features [...]
> </blockquote>
>
> [http://www-128.ibm.com/developerworks/library/x-xslt/]
>
> And it was proven to be Turing complete. Hovewer i agree with you, even
> focusing to certain kind of simple transformations for small docs (<
> 3Mb?), the language is too complex and convoluted.
>
>   
>>> I think that XSLT is much more difficult than other languages to
>>> learn. Others agree with me.
>>>       
>> In my experience, only those who come at it from a functional
>> programming perspective. Given that XSLT thrives when being
>> declarative, I see more people struggle with XSLT because they haven't
>> got a clue what they're doing, nor have they read what it says on the
>> tin.
>>
>>     
>>> <blockquote>
>>>       
>> ...
>>     
>>> <blockquote>
>>>       
>> ...
>>     
>>> <blockquote>
>>>       
>> ...
>>
>> This is silly. Why didn't you search for "XSLT rocks" instead?
>> -- http://www.43things.com/entries/view/417053
>> -- http://norman.rasmussen.co.za/45/xslt-transformations/
>>     
>
> What is silly? That you erased the links contradicing you? That Michael
> Kay said that "That means there is a steep learning curve and often a lot
> of frustration"?
>
> Of course if an average programmer changes the 'chip' from usual
> imperative to declarative programming and become familiar with the
> redundant angle notation, then (s)he perfectly may find easy XSLT.
>
> Well, also raw binary code may be easy with enough effort from our part...
>
>   
>> Obviously not all agree it sucks. I can also provide you with quotes
>> that says that once you do what you're supposed to do instead of doing
>> what you think you should do, XSLT is *the* XML tools of choice.
>>     
>
> Many other strongly disagree! I already cited a bit. Now i can introduce
> "Transcending the limits of DOM, SAX, and XSLT" article
>
> [http://www-128.ibm.com/developerworks/library/x-matters14.html]
>
> <blockquote>
> Aside from the somewhat annoying verboseness of XSLT, it is limited in its
> expressiveness -- the things you can say are expressed rather clearly (and
> functionally, not procedurally), but you quickly bump up against all the
> things that you simply cannot say in XSLT.
> </blockquote>
>
>   
>> I had a
>> somewhat steep learning curve with XSLT *because* I didn't
>> understand what it was meant to do. Once I actually read the darn
>> specification, I felt ashamed and embaressed, and went on to a
>> declarative and keyed happier lifestyle.
>>     
>
> That is good for you. I also worked in a text to MathML processor using
> XSLT and in a XSLT transformation from original (XML based) CanonML to
> (X)HTML for presentation. With a lot of effort you can do many things
> work, but real problem was in extension to scientific module. Even
> asumming only three lines of XSLT code for each scientific concept of
> language, i would be not glad with a 6000 lines XSLT only for the
> physicochemical module.
>
> Also lispers try to convince us that prefix notation is closer to mind's
> God, but still prefix notation lacks popularity in all programing
> languages i know.
>
>   
>>> I prefer to think that XSLT is just no popular when compared to other
>>> wb technologies. It appears to me more reasonable conclusion.
>>>       
>> Popularity has nothing to do with quality. DOM is probably used more
>> than XSLT. Would you use DOM or XSLT to squeeze goodness from an XML
>> tree? And more importantly, why?
>>     
>
> About quality i agree with with you.
>
> About DOM, well i think that often both approaches are complementary.
> 'Standard' DOM can be slower than XSLT, but inner method improves a lot
> of.
>
> I would are surprised if you are able to reproduce ASCIIMATH using XSLT
> for instance.
>
>   
>> Alexander
>> --
>> "Ultimately, all things are known because you want to believe you know."
>>                                                          - Frank Herbert
>> __ http://shelter.nu/ __________________________________________________
>>     
>
>
> Juan R.
>
> Center for CANONICAL |SCIENCE)
>
>
>
> !DSPAM:450012e1155341960693638!
>
>