[Docutils-develop] Re: How to add a tex-math role?

[Felix W. <Fel...@gm...>]

David Goodger wrote:

> A setting/option could be added to the HTML writer to turn off "name"
> attributes.

Too much bloat, IMO.

> As long as Netscape 4 is still in use, we can't drop the 'name'
> attribute.

We can.  Very few people are using it today, and current wisdom in
newsgroups seems to be "they can be glad if they see anything."  I mean,
who cares if NN4 users can use the table of contents or have to navigate
manually?  And after all, the changes we do *now* propagate very slowly
to users actually viewing the HTML pages.

I wouldn't mind waiting another year or two 'til we remove the
attribute, but only as long as it doesn't cause extra complications in
the code.

>> Which remembers me of the Docutils-internal id/name mess.
>
> <ahem> Excuse me?  I didn't realize there was a mess at all.

I found it quite difficult to understand; and I assumed it was really a
mess because I misread the to-do list entry about names and IDs.  Sorry.

> Docutils names are the human interface to IDs.

Now this makes things somewhat clearer.  Added to Element docstring; if
you know a better place to put it, feel free to move it around.

>> And the design seems to be quite some pain, given that we still
>> didn't manage to fix the mislocated-targets bug (at
>> <http://docutils.sourceforge.net/BUGS.html#mislocated-targets>).
>
> That bug is completely unrelated to the handling of names & IDs.

It's not.  We need multiple IDs and names per node to fix that, don't
we?  The current name/ID system doesn't allow that.

Do you think we could turn the current 'name' and 'id' attributes into
'names' and 'ids' list attributes?  (We get XML deserialization issues
then, but we have those issues already, because the backrefs attribute
is a list; so it doesn't matter for the moment.  It should be possible
to fix that by adding some magic to Element.)

If yes, what about auto-initializing them with []?  It would just
require additional support in Element.starttag and Element._dom_node, so
that not every element gets an additional "id='' name=''" in the
(pseudo) XML representation.

> It should be a relatively easy bug to fix; all it requires is time and
> will.

Most bugs require only time and will to fix...  :)

This one will require a fair amount of both, I guess.

-- 
When replying to my email address, please ensure
that the mail header contains 'Felix Wiemann'.

http://www.ososo.de/