Syd Bauman wrote:
> Hi Martin! I hope to get to see you in College Park, yes?
Actually I'm not going to be there -- too many looming deadlines, and no
travel budget :-)
>> <rendition xml:id="Transcription">
>> <label>Transcription de texte</label>
>> <code lang="text/css" rend="ellipse">color: #ff0000</code>
> This is *so* much better than <hi>, I can't even begin to describe
> it. But I don't understand the rend= attribute, here. The semantics
> of the <code> element above are:
> "Here is a snippet of text in a formal language known as
> 'text/css'; in the source document, this snippet was rendered
> [as|in] an ellipse"
I just have to get the ellipse, spiral, rectangle etc. in there somehow.
If you know a better way, let me know.
> or some such.
> I'm also not 100% sure why you'd prefer the above encoding to the
> more expected
> <rendition xml:id="Transcription" mimeType="text/css"
> n="Transcription de texte">color: #ff0000</rendition>
> The only major loss I see is that you can no longer state what
> natural language the string "Transcription de texte" is in, and it
> can't have characters outside of Unicode.
Actually, since "Transcription de texte" is XML markup, that won't work.
It could be:
Transcription de <emph>texte</emph>
so it has to appear as tag content.
>(Yes, you've also lost the
> "ellipse", but since I don't know what that means, it's not a loss
> that I can see :-)
Unfortunately, it's essential. Every category of annotation has a shape
associated with it, which is selected by the user, and serves to help
distinguish different categories of annotations on images where there
are many annotations. That's been a key part of the IMT since the beginning.
>> Looking at the range of tags now permissible in <rendition>, which
>> include everything from <bloc> (geo-political unit) to <depth>
>> (measurement across the spine) to <soCalled> ("scare quotes" etc.),
>> it seems almost perverse that the widely-used <hi> is not
>> available, but perhaps there is a convincing logic to it. Before I
>> rewrite all my code, though, I'd like to check that no-one has any
>> plans to make <code> unavailable in <rendition> in the future; I'd
>> like to avoid going through this again! If anyone is aware of any
>> upcoming changes to P5 that would affect the content model of
>> <rendition> (or for changing the attribute set available on
>> <code>), please let me know!
> First, allow me to commiserate about the bizarre children allowed in
> <rendition>. I personally would like to see a much shorter list. (In
> fact, your suggested use of <hi> convinces me completely! :-)
Oh curses. Now I suppose you're going to get rid of <code> as well.
> To my knowledge, there are no plans afoot to change the content model
> of <rendition> nor the attributes of <code>.
I'm trying to believe you, but in view of your comments above, I'm
> On the other hand, if you were going to limit the content of
> <rendition>, what would be the list of elements to which you would
> limit it?
I'd go back to the model that allowed <hi> in it (adding model.phrase?).
I think restriction of this kind is best left to customizations, which
can do a better job of it than the designers trying to guess what anyone
and everyone might want to put in their <rendition> tag.
>  That's not entirely true. There is another shortcoming of how the
> current system is intended to work. While it is easy to
> differentiate prose descriptions from formal descriptions *which
> have MIME types*, there is no mechanism to say what language a
> formal description is in if that language does not have a MIME
> type. I.e.:
> <rendition>Dark red against a putrid green
> <rendition mimeType="text/css">
> color: #660000;
> background: #33cc33;
> are both quite clear. But what if I have my own syntax for such
> foreground = C 16 M 86 Y 94 K 42.
> background = C 54 M 00 Y 75 K 00.
> there is no easy way (for the computer) to differentiate that
> from prose.
> I'm wondering whether folks
> a) like the <code> inside <rendition> idea, no change needed;
> b) think <rendition> should bear the lang= attribute in addition
> to (or instead of?) mimeType=;
> c) don't think this is a problem, no change needed.
Since <code> may be used elsewhere, I'd like to see both @lang and
@mimeType available on both <code> and <rendition>. Even if I'm using
CSS in <code>, I can't currently use @mimeType="text/css", which seems odd.
University of Victoria Humanities Computing and Media Centre
([log in to unmask])
Half-Baked Software, Inc.
([log in to unmask])
[log in to unmask]