Mariage

Clarifications:

1. So far, decent progress. I think I finally understand some of the basic mechanisms of xml, which is exciting, after having done it (intermittently, to be fair) since January!
2. New fun batch of issues: I'm struggling with if I can encode the borders of the text I'm working on. I've spent the last half hour on the TEI website and I'm starting to wonder if this is indeed possible. I found a "decoNote" that included borders as part of an "msDesc", but since I'm not working on a manuscript, I don't think this is appropriate. Also if I were to use this element, it has a number of obligatory component elements that aren't applicable seeing as I'm not working on a manuscript! So still working on that...
3. Still need to encode the stamps and make sure my foreign elements are correct and that they should not be "xml:lang"s instead.

From LSPW, looking at the two Sonnet texts on the site:

1. After the title page of both sonnets, there should be a blank page. In the 1621, there is one, but the blank page is missing from the 1609.
2. Italic font style was used to transcribe the "Epistre" part in the dedications of the 1609 text but the rest of the text shows up in normal text that is not stylized as it is in the 1621. However, in comparing the facsimiles, the font could still be considered to be in italic style all the way through. We should change this in the xml to reflect this.
3. In the 1609, many of the capital letters starting off a new paragraph are too far off in the left margin. These include lines 273, 301, 322, 343, 377, 432, 454, 477, 499, 648 in the xml.
4. The 1609 title "SATYRE MENIPPEE." (line 699 of the xml) where the actual sonnet begins is not centred and the word "SATYRE" should be above the word "MENIPPEE".
5. In the form work of the 1609 version, in the header there's a space after each letter to reflect the style as seen in the fascimile. For instance, instead of writing "SATYRE" we have "S A T Y R E". This should be done for the 1621 as well.
6. On page 19 and 21 of the 1609 (there are other pages as well), the phrases "LE SAN-GVIN." (lines 1239-1240) and "LA FLEG-MATI-QVE." (lines 1262-1264) appear on the left when they should be on the right.
7. If you look at page 18 of sonnet 1621 on the website, there is text on the left margin that shouldn't be there. There must be something wrong with the xml but I don't know how to fix it so I'll leave it to you.

LCC has now got quite a way into the markup of the Varin, so I've pushed it into the db to see what it looks like. It's pretty good all round, but there are some issues that need attention:

• The layout of the recto and verso pages is very different (margins are tiny on the binding side, and huge on the other). The current simple system for specifying the original page layout can't easily do this, so it needs to be elaborated. We can distinguish between recto and verso based on page number and/or page number position, and perhaps specify two different page layouts.
• There is a lot of quotation without quote marks in the original text, and LCC has made a nice, useful distinction between two different types, marked up with <quote> and <said>. Right now, <said> is not handled at all, and <quote> always results in the insertion of quote marks. What we really need is a way to mark <quote> tags which do not have quote marks in the original (perhaps @rend, or @type). Once we have this, we'll be able to do something quite neat: render the "original view" without quotes, and supply quotes in the "version continue".

Spent a lot of time tweaking XSLT and the markup of the Sonnet documents to get some output which is reader-friendly and represents the original texts somewhat. This is what I've done:

• Fixed many inconsistencies in the 1621 markup, where older markup formulations had been used instead of the ones we later settled on. This included lots of fixes for <fw> tags to make them consistent with the 1609, and with the Amboise texts.
• Added basic layout information to the heads of both Sonnet files to set page sizes etc. This will need more tweaking, though.
• Fixed some font sizing issues in the running headers in 1621. The headers in 1609 still look too small, but I need to check against the Amboise, which is the standard, to make sure I'm handling them correctly in the 1621 before I move on to the 1609.
• Fixed a distracting background colour for the <unclear> tag output.
• Added a comment ("KTO", for "Keep Tag Open") inside XHTML <div>s which occasionally end up empty in the output, otherwise the Cocoon serializer self-closes them, which screws up Firefox's DOM and makes the layout a mess.
• Pushed copies up onto the SAN so the editors will start from my fixed versions, and uploaded them into the db for testing.

Tweaked the XSLT as suggested in the previous post, and the Sonnet XHTML output documents now have no validation errors. That's removed over 2000 errors from the 1609 :-). The output is still a bit weird -- margins seem to keep incrementing down through the document -- but it'll be easier to track that down and figure it out now.

The two editions of the Sonnet present quite complicated rendering problems for our site, because their page structure and line-group structure overlap. Specifically, we have line-groups running over many pages, and so the conventional way we're handling verse lines for the shorter poems will simply not work (we're using <ul> and <li> elements, but the <fw> tags which appear in between lines at page boundaries are block-level elements and need to be rendered using XHTMKL <div> tags, which cannot appear between the <li> elements inside a single <ul>).

This needs to be carefully handled to avoid disrupting the display of the shorter, simpler poems already on the site. These are my preliminary thoughts on how we might do this:

• Detect the problem when matching <lg> and <l> elements. Specifically, note when <lg> has a child <fw>, or <li> has a parent <lg> with a child <fw>. It might even be better to detect this structure's existence anywhere in the document, because that should dictate that all line groups and lines be handled in the same way throughout (we don't want some handled one way and some another, because some accidentally happen not to contain <fw> tags).
  
  So we could set a single variable at the beginning of the rendering process, based on the existence of an <fw> tag anywhere in the <text> element which has a parent <lg>.
• Where this situation is detected, branch the handling of <lg> tags and <l> tags, so that <lg> becomes a <div class="lineGroup">, and <l> becomes a <span class="verseLine"> followed by <br />.
• Provide appropriate CSS so that these render correctly.

I'm going to start work on that today.

Meeting with CC, EM and LCC about summer work schedules, material for the DHSI workshop they'll be doing, etc. EM is still working steadily on the annotations; LCC will finish the Varin transcription any minute now, and will start work full-time on the markup next week -- I've set aside some time to get her started. More details soon...

PG hit a markup issue: the 1609 Sonnet document has two distinct works in it, and the second of them has as elaborate a title page as the first, occurring in the middle of the document. However, the <titlePage> tag only seems to be able to appear in a <front> tag, which cannot appear after a <body>. I think the obvious solution is shown here, and looks like this:

<text>
 <front/>
 <group>
  <text>
   <front/>
   <body/>
   <back/>
  </text>
  <text/>
 </group>
</text>

The frontispiece/title page material that applies to the whole document comes first, in the root text, and then the two component documents are enclosed in a <group> tag, each being a full <text> element.

All the 2004 symposium lectures and their media are now complete. Only four more years to go (2005-2008).

The two issues described here have now been fixed. There are still many issues with the two documents, some of which require PG and LSW to fix markup inconsistencies, but some of which (including many listed in PG's email of March 27) require intervention from me and/or decisions from CC. In particular, we must set the document page widths appropriately for the non-continuous rendering, in the <teiHeader>. I don't want to start on this till I have completed markup from PG and LSW, though; the 1609 doc is currently invalid and ill-formed, so it needs fixing anyway before I can upload recent changes.