[tei-council] Should Roma be doing this?

Martin Holmes mholmes at uvic.ca
Wed Feb 6 20:20:11 EST 2013


On 13-02-06 03:31 PM, Sebastian Rahtz wrote:
>
> On 6 Feb 2013, at 22:32, Martin Holmes <mholmes at uvic.ca> wrote:
>
>> I think the core problem here is that we've been using @type on <list>
>> for all these years when we should have been using @rend.
>
> yes and no. I can't shake my conviction that every _modern_ system does make
> that fundamental distinction between the ordered and unordered list. Yes of course
> we are about transcribing what we see, but those modern things didnt appear from nowhere.

But the same thing applies to italics, or drop-caps, doesn't it? As Syd 
says, lists are ordered and can't be otherwise; the issue is whether 
they're explicitly numbered, and that's a typographical rendition issue. 
I think the unfortunate issue of HTML ul/ol versus the CSS that should 
do the job (list-style-type), but which only came later, has confused 
our thinking on this. If there were a true difference in type, wouldn't 
TEI have created two separate elements?

>> Doesn't this
>> sound more sane?
>>
>> @rend="bulleted"
>> @rend="numbered"
>> @rend="simple"
> well, sort of. I've always disliked the over-specific "bulleted", and I've never understood
> "simple" but I accept thats my problem :-{

At least "bulleted" is a clear description of the rendering of the list.

>> The only value of @type that really looks like a type to me is "gloss",
>> but this is used, as far as I understand it, when there is a <label>
>> preceding the <item>, and in that case the presence of <label> is
>> sufficient to determine the output appearance.
> yes, thats true. if there are label children, its a gloss list.
>>
>> So I would recommend:
>>
>> 1. Changing our recommendation so that we provide suggested values for
>> @rend, not @type.
>
> ok, I suppose
>
>> 2. Adding a suggestion that using @style with CSS will provide a useful
>> range of precise options that cover most cases.
>>
> or better, @rendition, in fact. @rendition="#romannumbered" is a more useful
> scenario

How about <list style="list-style-type:upper-roman;">?

>> 3. Leaving @type on <list>, for backward compatibility, but providing no
>> suggested values for it.
> not suggested values, but with an example? can we think of one?

<list type="instructions">

<list type="recipe">

<list type="shopping">

These are all actual types of list; any of them might be rendered with 
bullets or numbers, without changing their type.

>> This will cause Sebastian considerable stylesheet-related pain, I know,
>> so apologies for that.
>
>
> yes, I haven't the faintest idea how to know whether to make  <ul> or <ol>.

Yes, damn it, that's the heart of the problem. HTML was wrong to have 
two separate tags. But if you create a <ul> and make it list-style-type: 
upper-roman, you'll get roman numerals, and if you give an <ol> 
list-style-type: disc, you'll get a bulleted list. If we believe that 
all lists are inherently ordered, we could just use <ol> in the output 
and let CSS do the styling.

> I suggest that this is such a big bucket of worms that it needs public discussion.
> @type on <list> seems so well established in both instances and tools that
> I can't help feeling we may provoke howls of protest across the seven seas.

Indeed. But we don't have to break backwards compatibility; you could 
still detect the old values in the stylesheets and retain the old behaviour.

Cheers,
Martin

> --
> Sebastian Rahtz
> http://www.justgiving.com/SebastianRahtz
> Director (Research Support) of Academic IT Services
> University of Oxford IT Services
> 13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431
>

-- 
Martin Holmes
mholmes at uvic.ca
UVic Humanities Computing and Media Centre


More information about the tei-council mailing list