At 03:31 PM 1/18/01 -0500, Jeffry Smith wrote:
>Leonard Rosenthol said:
>> At 8:39 AM -0500 1/18/01, Jeffry Smith wrote:
>> >I'm a little lost here. I thought a section was like a <div> in HTML.
>> >Example: I have a part of a document that contains a bunch of wide
tables,
>> >some images, and footnotes. I decide to print that part in landscape
mode,
>> >with the rest of the document in portrait. That section would be a
<section>.
>>
>> The example is more accurate than the HTML comparison, but
>> yes. A section is an area of a document that uses a different layout
>> (number of columns, page size, etc.) than previous/following sections.
>
>Hm. Don't see how a table necessarily fits into this. Especially as we
>implement things like floating tables (or floating frames, the bottom part
of
>my proposal), that allow the renderer to shift things around within
>constraints to fit the final document.
>>
>>
>> > Can sections nest?
>>
>> I don't know about the current implementation, but I believe
>> that they should be able to, yes. Example, you start a section for
>> landscape mode, then inside of that is a section with a column change.
>
>If a table is a section, it will need to be able to nest. This may be
getting
>into too much overloading of section. I'll start examining the code.
Jeff,
Thanks for taking the lead on specing out a tables design. It's high time
that someone started diving into all the complexities of this mess. :-)
I'm not even going to pretend that I'm up to speed on this entire
discussion, but I want to confirm your initial intuition about sections.
- Tables are not sections.
- Sections are not tables.
- Tables are not columns.
- Sections are not pages, either.
A document is a linear series of logical sections, each of which can have
varying properties such as page size, orientation, column layout, etc. A
section can break to a new page, or it can break to a different layout on
the same page.
Headers and footers are currently implemented as sections, too.
I suspect that you'll want tables (and their constituent cells) to be a more
complex kind of container. I haven't though this through, but chance are
we'll need to revise both the logical and physical hierarchies in the
formatter. For more details, see the (dated) comments here:
abi/src/text/fmt/xp/fl_DocLayout.h
I suspect that this kind of comment is exactly what we most need in Doxygen.
Paul
