Leaving aside the <fw>, there appears to be some support for having multiple <pb/> per page, (at least) one for each text stream. I doubt that the standard stylesheets behave well with this.
In some scripts I write, I have trouble with notes and figure descriptions and the like, which I have thought of as "misplaced text" (text that occurs in positions in the text stream of the body that do not correspond to their position on the page).
I think we should see the text as consisting of a number of coordinated text streams. Ideally, these should be marked up in isolation: the standard TEI approach, to mark up a text consisting of multiple text streams as if it were one, is perhaps not the best, leading to the common workarounds described by Peter.
The note text stream is coordinated with the body text stream, and when it is entered and has broken across the page and terminates, it rejoins the body again, without the need for an (event-based) extra <pb/>, I think.
On 14 Feb 2014, at 16:02, Paul Schaffner <[log in to unmask]> wrote:
> On Fri, Feb 14, 2014, at 9:09, Jens Østergaard Petersen wrote:
>> Another milestone puzzle I have is this:
>> Say your text has a body with several pages, some formework which spreads
>> over facing pages, and - to boot - a footnote which starts on one page
>> and ends on another. How many <pb/> would you use?
> I think that each <note> that extends onto a second
> page should contain two <pb/> tags, one in the middle
> of the note to indicate the movement to the next page,
> and one at the end of the note to indicate the movement
> *back* to the original page.
> On the theory that <fw> is a sort of note, i.e, it
> captures something in the original that is outside
> of the main text stream, then I suppose that if I were being
> punctilious, I would treat <fw> the same way.
> This sort of thing is one reason that we (for the
> most part) do not capture <fw> material at all. When
> we do, we tend to bind the <fw> very closely to the
> <pb>, and in your example would probably instead
> capture the forme werk in two <fw> tags, one for
> each page, rather than allowing a <pb/> to appear
> within a single <fw>. Or we would quietly ignore
> the fact that the <fw> spreads across a page opening
> and treat it as if it appeared in its entirety
> on a single page (or, better, repeat it on both pages).
> (As for run-over notes, we mostly also ignore the advice
> above, and treat the note as if it all appeared on
> the page on which it started, But we certainly do
> take the multiple-pb approach to other features that
> lead the reader to turn the page over, then back,
> sometimes repeatedly, e.g. parallel-column texts.
> A <pb/> for us tags an event -- that of turning
> the page, in either or any direction, however many
> times are necessary.)
> (And as for your earlier question -- for reasons of
> our own, we in fact do *not* allow <pb/> to appear directly
> in <body> <back> or <front>. Since we retrieve text
> by the <div>, no <pb> is allowed to occur except
> within a <div>. I think this change to our schema
> was suggested by Michael Popham, about twelve years
> ago!) This does, however, still allow <pb/> in element-only
> contexts. But the same could be said for allowing it
> (say) between the items of a list, or between the lines
> of a stanza.
>> On 14 Feb 2014, at 11:59, Jens Østergaard Petersen <[log in to unmask]>
>>> In all the examples given in this thread, <lb/> and its siblings (imagined as well as real) have been placed in the text stream, that is, as marking transitions in the text of an element which with mixed contents.
>>> One thing that worries me a little is that the schema allows me to place (the empty) milestone elements among the element-only top-level elements such as <text>, <body>, and <div>, that is, outside the text stream. This is very common with <pb/>, but it is also found with <lb/> (see e.g <http://www.deutschestextarchiv.de/book/download_xml/dilthey_geisteswissenschaften_1883>).
>>> The reason I write this is that I have my own problems related to milestones and the question "which elements are block-level elements (and how do I retrieve them)", and this may not be relevant to the present discussion, but I still ask myself, how can a line break occur where there is no text to form lines? does a page break at the divisions we see in the text or at the point where the characters in one page end and those in the next page begins?
>>> A related question is why anchors are allowed to stand outside the text stream, as children or siblings of elements that could easily have an xml:id themselves?
>>> Sorry for sidetracking the white-space discussion!
>>> On 10 Feb 2014, at 15:48, Sebastian Rahtz <[log in to unmask]> wrote:
>>>> On 10 Feb 2014, at 14:39, Scott Derrick <[log in to unmask]> wrote:
>>>>> Having a tag or set of tags that absolutely encapsulates a line of text would eliminate half the kludgy(my words) code dealing with it and the output would be more consistent and correct.
>>>> there are (so they say!) many ways to skin cats. including:
>>>> * use semantic tags like <l> or <seg>
>>>> * use an authoring schema which makes life easier for humans to read XML (o nefas)
>>>> * use another markup language, and keep TEI for non-human interchange
>>>> * use milestones and anchors
>>>> * use transcription surface/zone/line markup
>>>> so you have to decide what’s best for a particular project
>>>> If we _did_ have the tag for line-end, as in:
>>>> <p> and so he said
>>>> <sl/>this is the way the world ends<el/>
>>>> <sl/>not with a bang but a whimper<el/>
>>>> in a quiet voice</p>
>>>> I wonder what you think the whitespace between <el/> and <sl/> means?
>>>> Sebastian Rahtz
>>>> Director (Research) of Academic IT
>>>> University of Oxford IT Services
>>>> 13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431
>>>> Não sou nada.
>>>> Nunca serei nada.
>>>> Não posso querer ser nada.
>>>> À parte isso, tenho em mim todos os sonhos do mundo.
> Paul Schaffner Digital Library Production Service
> [log in to unmask] | http://www.umich.edu/~pfs/