Re: CSS discussion

On 8/7/07, Abel Rionda <abel.rionda@fundacionctic.org> wrote:
>
>
> TODO Line Number
>
> Adding a line number property to each serialized CSS property (This can be
> done during serialization).
>

I am worried this is not trivial. Every element will have a line number?

TODO: Deal with short hand properties (eg font into font-size,
> font-family….)
>
> TODO: Parsing out the values defined in properties into usable values
>

These too are a lot of work.

*Comparison between CSS in XML and Regular Expressions*
>
> * *
>
> It would be very desirable if you can complete these issues in order to
> make a comparison.**
>
> * *
>
>    - Structure of moki
>
> We are not sure how is the final moki structure in this way. Can you
> include a sketch of stylesheets information moki?
>

There would be no structured representation of CSS in the moki. The moki
records the retrieval of external stylesheets, and of course of styles in
the primary document. Therefore all the text of the CSS is readily available
without any additional representation.



>
>    - Structure of Tests
>
> Currently we have two CSS tests (Measures and StyleSheetSupport) running
> made in a pure XSLT way based on CSS serialization
>
> How would be a skeleton of any of CSS test with Regular Expressions?
>

I don't understand, at some level you need to find and parse out things like
"50px" into "50" and "px". You note this above. Regexes are a good way to do
it but I'm open to anything else. Therefore this is required no matter what.
MEASURES is merely a matter of looking for certain patterns in text that
appears in certain elements in the primary document, or in external text CSS
files. STYLE_SHEETS_SUPPORT is equally trivial.



>    - Line number
>
> How is reported the line number with this approach?
>

Very easily. We already have line numbers in the primary document, which
covers any CSS in the primary document.
External stylesheets are just text files. They can be processed line by line
looking for the things we want. Line number is trivially available that way.


Serializing CSS into XML is interesting, but I do not see it meeting any
need here. I do not favor completing it now, and since it is incomplete, I
favor removing it from the code base. I am still not clear on the arguments
for this approach? It is strictly more work. It does not make anything
easier that I can see.