Map Of London

Since we still don't have an actual map image to work with for the new interactive map, I'm having to reconstruct the old map interface in the redesign style, since we're going to be launching with it. I've done everything except the JavaScript bit so far, so tiles are loading and navigation arrows are working, and the drop-down selector is using the new document type taxonomy instead of the old @type attribute.

Meanwhile, I've also been thinking of a "slippy" version of this crude interface, and to that end, I've pieced together all of the original tiles into a single large map. This took quite a while, but it will enable me to create a sort of hybrid where you can slide the map around, and it shouldn't be too hard to figure out how to offset the coordinates for each individual tile and get the places to show up in more or less the right location. We shall see.

Today:

• Reworked the taxonomy a bit in consultation with TL (removing the Editorial categories because they didn't seem to mean anything, among other things).
• Restructured the taxonomy to implement nesting, so we have a hierarchical structure.
• Added new categories for all the different location types.
• Wrote code to generate the correct catRef elements in all the locations documents.
• Generated a new section for the ODD file encoding the categories and built a new schema.
• Tweaked and updated the textClass in all documents in the db to conform with the schema.
• Rewrote the document type listings, so that they now require only the lowest level of a nested category set to be present in order to function (which eliminates redundancy in the textClass), and also so that different document types are displayed in different tables with different columns to suit their type. I also added a feature that lists the subcategories of a category, acting as a sort of landing page.
• Added XQuery to generated a nested list of the entire taxonomy so you can see it at a glance.
• Added a documentation file explaining how it works, and what encoders have to do.

Pulled the little rose out of one of the design mockups and turned it into a favicon for the redesign site. I was tired of seeing that tower, which has nothing to do with the new design.

Based on our rich <respStmt> system, I've now enhanced the display of information about people considerably. It now includes a list of documents the person authored, a list of roles they have taken in documents, and a list of documents which mention the person.

The popup display also now includes a little link next to the name, which takes you to a full-page view of the same information. This is something we've talked about for a while; it will enable everyone who participates in the project to provide a single URL where their contributions are enumerated, and it also gives anyone wanting to link to an entry in our personography a URL to link to.

There are still a lot of old documents with <byline> elements that need to be converted to <respStmt>s, but all our newer documents should have proper attribution.

I've made a number of changes to the schema to tighten up our markup practices:

• The four elements of the teiHeader are now mandatory (fileDesc, profileDesc, encodingDesc, and revisionDesc), and must be supplied in that order.
• profileDesc must contain textClass with at least one catRef in it.
• catRef/@target values are now constrained by the schema to conform with the list in our taxonomy. The schema will suggest the values and provide explanations of them.

I've updated hundreds of documents to make them conform with the new restrictions.

Rewrote the XML id index page as a full site index for regular users, and also rewrote the JavaScript that handles table sorting so that it shows more obviously that sort is available, with black triangles, and shows the active sort column and direction with a red triangle. Note: if at some point we need to define a collation, there's a lovely solution here: http://stackoverflow.com/questions/3630645/how-to-compare-utf-8-strings-in-javascript.

Implemented some fixes to citation styles.

I have it half-solved; I may need to refactor things a bit to make the menu stay at full size (assuming that's what we want) when you take your mouse off it.

We really need to do some accessibility testing for the main menus too.

Added a RIS export option to the citation popup, providing a text file. This should be served with the correct mime type, but until we rebuild eXist we can't incorporate the mime type, so I'll need to remember to do that next time we rebuild, and then edit ris.xql.

I then added a rudimentary COinS span to the box (this is an invisible span that hides metadata in its title attribute). This looks like it would be a good idea, and Zotero can apparently read it, but it turns out that the spec is so sparse that it's hopeless; you can only do books and journal articles, and there's no way to include an editor. No use at all for proper work. I think we'll need to work on encoding bibliographic data into the HTML meta tags, or as RDF XML, but I need to find out what tools are able to read this appropriately. One other possible model is the Dublin Core we put in to the meta tags of the Despatches pages.

I've reconfigured things so that in the redesign, the menus are created based on nested lists in this file:

/db/data/boilerplate/site_menu.xml

There are two lists in there, one for the main menu and one for the submenu. They're both set up right now more or less according to what emerged from yesterday's discussions of KMF's Powerpoint slides.

The editorial team now has the power to change the menus, so they can try things out without needing me to implement them. The structure is pretty straightforward -- each menu or submenu is a <list>, and each item in the menu is an <item> containing a <ref>. Linking is done in the usual MoEML way with mol: prefixes.

The menu on the old site is controlled in a similar way using content in the boilerplate.xml file. We need to leave that alone for the moment, but once we've moved over to the redesign we can remove that content.