Re: Idea for simpler lists and tables

[Jim Campbell <jwcampbell gmail com>]

Hi All,

On Mon, Jan 4, 2010 at 11:00 PM, Shaun McCance <shaunm gnome org> wrote:

Hi folks,

In Mallard, every element unambiguously takes either block or
inline content.  In other words, you can't do something like
put a paragraph inside a paragraph.  This was an intentional
design decision from the start, and it considerably simplifies
a lot of processing.

One side effect of this is that, in order to allow different
block elements in list items, we have to enforce that list
items always take block content.  So we have to do this:

<list>
 <item><p>One</p></item>
 <item><p>Two</p></item>
 <item><p>Three</p></item>
</list>

The same applies for table cells.

We've discussed ways of dealing with this little annoyance
in the past.  One suggestion was to introduce a different
list item marker that takes inline content instead.  This
would work, but I'm very cautious about adding elements.

So I had an idea.  What if we just allowed block elements
to stand on their own, without an item element?

<list>
 <p>One</p>
 <p>Two</p>
 <p>Three</p>
</list>

Any block element that's directly inside a list would be
interpreted as if it were wrapped in an item element.  Of
course, you would still be able to use the item element
if you need to wrap multiple blocks elements in one list
item.  You could even mix and match in a single list.

<list>
 <p>One</p>
 <p>Two</p>
 <item>
   <p>Three</p>
   <p>i.e. 3</p>
 </item>
</list>

The same would apply for item elements in a steps list,
but not for item elements in terms lists or tree lists.
(The processing model for those would make it stupidly
difficult to deal with.)  And we could possibly do the
same thing for table cells, though I'd need to look more
thoroughly into how much that complicates things.

For basic lists and steps lists, though, processing this
is actually very simple in any tools I'm familiar with.
I don't think it would significantly complicate anything
on the code side.  And it saves everybody a lot of typing.

My question is, does it make sense to you as writers?
It certainly seems much easier once you understand it,
but does it boggle you at first glance?  I want to make
things as simple as possible, but sometimes too simple
can just be confusing.

Thoughts?

Initially, looking at something like this seemed pretty cumbersome to me:

<list>
  <item><p>One</p></item>
  <item><p>Two</p></item>
  <item><p>Three</p></item>
</list>

In practice, though, it hasn't been much of a hindrance to me.  Once I knew that this is how I set up a list, it has been easy to remember, and easy enough to use.  I normally just copy and paste from one list to another as a starting point anyway.

You reference lists here, but steps are done in the same way.  If you made this particular change, would <step> (or other such syntax elements) also be affected?

Part of me thinks, "Will this cause problems with xslt processing or other stuff that we might be considering yet?" but other part thinks that if Mallard's goal is to make authoring simpler (more simple?) . . .  such a change would be good.

Jim