[sword-devel] CSS Again... (was Re: Rendering added words for languages that don't use italics?)

[Greg Hellings]

On Tue, Apr 19, 2011 at 9:35 AM, DM Smith <dmsmith at crosswire.org> wrote:
> In principle, I like the idea of CSS, but I think there are difficulties
> with CSS. It presumes the elements and structure of what is being styled and
> that the display can handle it.

Are these problems greater than our current solution of "Sorry, nope,
you can't style _anything_ because there are some niche cases where
styling might be bad"?

>
> CSS is a container model, but osis2mod unwinds the Book/Section/Paragraph
> structure of a OSIS input into marker elements. How would CSS be written to
> address this? Would it be written for the authored OSIS or the transformed
> OSIS?

Since CSS is a display technology and SWORD does not uses OSIS for
display, clearly this question does not even apply.

>
> SWORD has several renderers, HTML, RTF, plain text,.... How would CSS apply
> in the delivery context of RTF?

Can we stop even bringing up RTF? BibleCS's choice to use RTF means
that it will always be limited to the problems inherent in that
technology, just as every other front end has chosen HTML and is
limited by that.  This is one of the niche cases I referred to above.
"Well, BibleCS wouldn't be able to employ this SGML-based technology,
so let's just ignore it and keep pooh-pooh-ing it."

>
> For example, JSword transforms ThML, GBF, TEI and Plaintext into OSIS
> (augmented with SWORD's usage of TEI). Even OSIS and TEI undergo minor
> transformations. Then OSIS is converted into HTML. To what would the CSS
> apply?

To the HTML, of course. CSS can be applied directly to OSIS, but that
would require JSword to write a massive amount of CSS to give styling
to every OSIS element (just like it currently has an XSL file that
transforms every OSIS element). So long as the application is using
HTML, then the stylesheet would be applied directly to the HTML.

>
> The ability of a frontend to use CSS might be a problem. If I remember
> correctly, Xiphos was unable to use CSS. JSword/BD cannot at this time use
> an external stylesheet. Also JSword uses Java's built in HTML renderer and
> it is severely crippled and the HTML it requires is ancient.

Xiphos currently uses two different display technologies: gtkmozembed,
which is the Mozilla/Firefox/Gecko rendering engine. Where that is not
available (Windows) it uses gtkhtml. This makes the Windows build in
the same boat as JSword/BD at present. I am working with Karl, Terry
and Matthew to update to gtkwebkit for all platforms - both because
Mozilla is dropping support for embedding and because I want to see
proper support on Windows for Xiphos.

Of course, this line of argument is the same as digging up RTF.
"Because the widget I have does not understand this 14 year old,
fundamental web technology, clearly we should not support it."  BD is
limited by its choice of display technologies just like Xiphos is.
Xiphos has chosen to update its display technologies. BD would
probably benefit from opting to do so as well. Additionally, with the
CSS file on-disk, BD could simply insert the file directly into the
rendered output, removing the need for the stylesheet to be external.
I believe BPBible up until recently has used the wxHtml widget which
is also limited. Correct me if I'm wrong, but I believe they are
opting to update to a more functional widget. BibleCS, well.....

>
> I think that if someone had the initiative to write generalized code that
> would apply CSS to an OSIS module, use it successfully in a front-end, and
> submit it for inclusion into SWORD that it might happen. There are a lot of
> us that are quite willing to discuss an idea, but there are very few that
> will actually implement. Those that do implement prioritize what is
> important to them.

I had support in BibleTime for this already in place when I brought it
up last time. All that was needed was an agreement on a conf-file
entry name by SWORD.  But the idea of using CSS was summarily rejected
by the list. Similarly, Jaak gave a "Not on my watch" response to
enabling the technology in BibleTime should SWORD reach an agreement
on it. After all, then BibleTime would not be able to maintain
absolute dominance over every pixel of display. Never mind that it
already does not maintain this by virtue of allowing ThML-sourced
documents to have CSS styling tags on them and allowing external CSS
could actually reduce the impact of this loss of control.

So I went to Xiphos and Karl was not impressed with the idea. Digging
further turned up the fact that gtkhtml has about as good support for
CSS as Java does. So I offered to help with the gtkwebkit move for a
multiplicity of reasons, one being that hopefully Karl will consider
adding support for external CSS files to Xiphos after the rendering
engine supports them.  But if SWORD simply shoots them down again, for
virtue of "That's very nice but MY rendering widget can't handle CSS"
like happened before...

Part of having a choice of front-ends is the fact that some are
limited by their choices in technology. Perhaps BD has to sacrifice
support for external CSS to be able to run everywhere Java exists.
Perhaps BibleCS has to give up support for CSS because it chooses to
stick with RTF for some reason. Does that mean that Xiphos, BibleTime,
MacSword, BPBible, PocketSword, AndBible and whoever else chooses to
use a modern HTML widget should be so limited?

--Greg