I think I'm repeating things that have already been said, but here it
is in a different flavor.

For most of the documents I've seen, rectangles aren't particularly
interesting - this may be a result of spending too much time with
artists looking at illustrations or Picasso's manuscripts. Given that,
if the feeling is that TEI is fine for rectangles, but for everything
else we should just use SVG, why not use SVG for rectangles as well?

That line of reasoning seems to go well with the sentiment that TEI
documents should link to SVG based descriptions of
manuscript/facsimile images.

That said though, the idea makes me a bit uncomfortable. We're not
just talking about how to represent vector graphics, we're talking
about how to encode texts (albeit images of texts). This seems more
TEI-like than SVG-like even if we end up using the vocabulary provided
by SVG. While we may not need to reinvent the vocabulary, it seems
that the TEI is the right place for describing and educating scholars
about how to use that vocabulary - that is, of course, to the extent
that "go use this tool" isn't education enough.

That has all been said, in one form or another, just my two cents on the matter.

Neal

On Tue, Sep 9, 2008 at 5:30 PM, Neal Audenaert <[log in to unmask]> wrote:
> Hi everyone,
>
> I'm a bit new to the list, and haven't had time to look over the TEI
> and SVG in great detail, so it's not entirely clear what types of
> things are possible and what aren't, but I figured I pass along a
> couple of types of problems we are thinking about.
>
> We are looking at a couple of interesting cases, one of which is
> non-geometric regions - for example, I would like to attach an
> annotation to all text that has been underlined in red. I believe that
> was mentioned with the possibility of creating several annotations and
> grouping them with a <g> attribute which seems to work just fine.
>
> The other type of problem would be to annotate some part of a region
> but not others. For example, I might want to annotate a stain
> underneath a text, but not the text itself. Or I might want to mark
> scribal annotations in red ink that overlap both the main text on the
> page and annotations in a different color of ink. That doesn't strike
> me as something that should be particularly hard to do, but I
> certainly agree with the prior comments that we will need guidelines
> that are a bit more involved than "use SVG when rectangles aren't
> enough."
>
> Neal Audenaert
>
> On Tue, Sep 9, 2008 at 11:52 AM, Martin Holmes <[log in to unmask]> wrote:
>> Hi folks,
>>
>> Hugh Cayless wrote:
>>> I agree with this, but "SVG" by itself can't be the whole answer.  It
>>> would be useful  to produce recommendations and develop practices for
>>> handling complex linkages using SVG (and/or other technologies).
>>
>> I'm just wondering what features we could imagine that _couldn't_ be handled
>> by a combination of TEI and SVG. What did you have in mind? SVG is a pretty
>> extensive schema for representing graphical and geometric structures.
>>
>> Personally, I think the FACS component in TEI strikes a very good balance
>> between simplicity and power. It provides enough flexibility for most
>> straightforward digital editions of conventional texts, and yet it's not
>> difficult to learn. I'm not convinced that there are a great many situations
>> where non-rectangular zones are really needed, given that rectangular zones
>> can overlap. For instance, in the relatively common case of lines of text
>> written at a slant, while it's true that any given line might be most
>> precisely captured by a non-rectangular shape, it's also true that there is
>> a larger rectangle that captures the whole line quite effectively. That
>> rectangle may overlap with others capturing other lines, of course, but I
>> don't really see that as a problem (although it may present slightly
>> interesting challenges for GUI designers). Most graphical applications deal
>> primarily with rectangles, and even those which allow non-rectangular
>> objects or layers are probably handling them as rectangles at a lower level.
>>
>> I think of TEI as primarily a descriptive and analytic way of handling a
>> document, rather than a representational tool. For that reason, it seems to
>> me to be enough to state (through TEI XML) that a specific piece of text is
>> written in a spiral, and perhaps link to an image of it; it's not the job of
>> TEI somehow to represent its spiral structure through vector geometry or
>> paths.
>>
>> Cheers,
>> Martin
>>
>> Hugh Cayless wrote:
>>>
>>> I agree with this, but "SVG" by itself can't be the whole answer.  It
>>> would be useful  to produce recommendations and develop practices for
>>> handling complex linkages using SVG (and/or other technologies).
>>>
>>> Hugh
>>>
>>> /**
>>>  * Hugh A. Cayless, Ph.D
>>>  * Head, Research & Development Group
>>>  * Carolina Digital Library and Archives
>>>  * UNC Chapel Hill
>>>  * [log in to unmask]
>>>  */
>>>
>>>
>>>
>>>
>>>
>>>
>>> On Sep 9, 2008, at 10:09 AM, Syd Bauman wrote:
>>>
>>>> One way to look at this is that the purpose of the facsimile tagset is
>>>> to make easy things easy. I.e., to allow users who have relatively
>>>> simple requirements to meet them without delving into SVG.
>>>>
>>>> From this point of view, it might be argued that the TEI facsimile
>>>> mechanism is already too complicated.
>>>>
>>>> If one has a complicated task (like multiple non-rectangular zones
>>>> with lines of text that are not parallel, e.g.), one is probably going
>>>> to have to do complicated things. We already have a complicated way of
>>>> doing this -- SVG -- is there much to gain in inventing another?
>>>>
>>>>
>>>>> I'll be presenting a poster at the TEIMM in November on the digital
>>>>> facsimiles tagset and one thing I'd like to look at is changes that
>>>>> could be made to improve it. There are two real weaknesses that I
>>>>> see. First, that we are limited to rectangular bounding boxes
>>>>> (unless we use SVG), and second that there is no clear way to
>>>>> account for non-horizontal text (text written sideways, upside down,
>>>>> diagonally, in a spiral, etc.). These two issues overlap - a
>>>>> rectangular box around text written diagonally across a page will
>>>>> include a lot of unnecessary information.
>>>>>
>>>>> Have you all on the list come across other things that you would
>>>>> like to be able to do with the digital facsimiles tagset, but find
>>>>> that you can't? If so, how have you dealt with it and what changes
>>>>> to the TEI would you suggest?
>>>>>
>>>>> Thanks for any suggestions!
>>>
>>
>> --
>> Martin Holmes
>> University of Victoria Humanities Computing and Media Centre
>> ([log in to unmask])
>> Half-Baked Software, Inc.
>> ([log in to unmask])
>> [log in to unmask]
>>
>