W3C home > Mailing lists > Public > public-ppl@w3.org > February 2013

Re: Customer requirement, a critque

From: Patrick Gundlach <gundlach@speedata.de>
Date: Sun, 24 Feb 2013 15:26:19 +0100
Message-Id: <2EA2504A-0CF2-42A6-9F04-0B3A13A2DF5A@speedata.de>
To: xsl-fo Community Group <public-ppl@w3.org>

>>> Rather than just go in an modify them, comments on the requirements.
>>> http://www.w3.org/community/ppl/wiki/CustomerRequirements
>>> 
>>> 3. -1 talking about products, surely we should be talking of
>>> fo:blocks? Like it. Just a bigger version of 1?
>>>  Please fit [id range perhaps?] or the contents of a
>>> block-container, in n pages (please add 'starting on
>>>  new page' as an option)
>> 
>> I am not sure why you use the term "fo:blocks". As far as I know, this is not a special FO community. The Customer requirements are independent of a special technology.
> 
> I relate requirements to the xsl-fo wd. Hence requirements are for
> implementation against that document
> or a later version of it.

I think that most of the requirements are not solvable with any current or future version of XSL-FO, unless XSL-FO changes in a dramatic way. Thus I see no point in comparing the requirements to XSL-FO :)

>> Its not related to #1. There are arbitrary rules a customer can state about how to reach that goal. Implementing all these rules is impossible in a formatting engine, because these rules are highly dependent on the situation.
> 
> Eventually they will come down to an fo specification for pagination
> in a formatter?

Absolutely impossible. I have never come across two customers that have the same optimization strategies. Implementing all these n strategies into the renderer would still miss the case n+1 where there is yet another strategy.

> That is my target, it seems we differ somewhere Patrick?

My target is to tell this group what my problems are and how I've solved them. If this leads to someone else picking up this idea would be great, but not a necessity for me. On the other hand, if there is more than one (=me) person liking the approach, we can discuss if this is good enough to standardize (or formulate or whatever).


>>>  IMHO the formatter needs a get out clause? It is known that some
>>> languages (German?) are incredibly
>>> verbose compared to others. Is there any language for expressing
>>> max-white-space? I.e. if the ws
>>> gets more than x% then just do your best, or something along these
>>> lines? How else to provide
>>> the get out for the formatter, or would you want horrible big white
>>> space?
>> 
>> If a language A takes up 5 lines and another language B takes up 20 lines, then yes, in this case I want the whitespace to be 15 lines.
> 
> So what would you want to happen when the shorter language results in
> 'too much' white space?
> In your example, 75% white space? That seems impractical to me.

It doesn't matter if it sounds impractical to me or to you ;-) - It is a customer requirement that saves lots of $$$$ (or better: ). The customer needs (and has) a XML to PDF workflow that meets the requirement. In reality there is never a 75% whitespace but much smaller (10% or so).


>> The goal is to make all color (images) in all product catalogs (n languages) exactly in the same space. But it should not contain unnecessary whitespace. So the "longest text" products determine the positioning of the color images. See my two pdf files I've uploaded to the wiki.
>> 
>>> An alternative could be
>>> expressed as 'fit these items in the same number of pages' in all
>>> languages?
>> 
>> No. See the constraint above (color images on the same space)
> 
> OK, I'll disagree with this requirement, on the grounds that the
> result would be bad typography.

No, the example might be bad, bit it's an actual requirement in a current project and it really looks great in that specific example.


>>> 5. I don't understand this. What images? Provided by whom? Would an
>>> easier one be to simply
>>> say, where a container has n-columns, to require column-balancing?,
>>> i.e. leave ws at the end of
>>> the two columns, in equal quantity?
>> 
>> I've uploaded another pdf for that.
> 
> Where please?  URL?

On the wiki page I've inserted a link to the pdf: See http://www.w3.org/community/ppl/wiki/images/7/76/Images-in-columns.pdf

>>> 6. Is this a corner case Tony? I don't know. I can see the use where
>>> 'which page this block goes on'
>>> does not matter. IMHO your example is too specific?
>> 
>> It is a very specific customer requirement (which this page is about). I am not sure how to generalize that.
> 
> Do you think such a specific use case is applicable for a general
> specification such as XSL-FO?

Yes. Absolutely. The system should give the building blocks to be able to do 99% of all the requirements we collect on the wiki page.

Again, I think the XSL-FO road is a dead end. I do see its usefulness, but none of any more than trivial day to day documents can be achieved with XSL-FO unless you wind down the demands on the resulting document. 

See for example:

http://www.erco.com/products/download/others/downloaddaten-3992/en/dltut-1.php?aktion=_startseite&sprache=en&dir=10_catalogues/10_catalogue

This catalog is produced with such a system (not XSL-FO). The demands of the customer are very high. 

Patrick
Received on Sunday, 24 February 2013 14:26:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 24 February 2013 14:26:51 GMT