Map Of London

Tentative launch date is Monday 9th (meaning I'll do the reconfiguration of the webapp on that day). These are tasks that must be accomplished before then:

• The page-based views must be working for the Mayoral shows and other documents with pbs.
• We must have a policy and/or GUI controls enabling the choice of page-based display for the user (or defaulting to it for e.g. small-format devices).
• The mayoral Editorial Declaration document (info/editorialDecl_mayoral.xml) should be supplied with section ids, and the Mayoral Show documents should import all or most of its contents into their headers via XInclude.

These are the tasks for the launch day:

• Make a complete backup of the exist db.
• Make a static copy of /data/ called /data_old/.
• Point all XQuery in /site/ to the /data_old/ directory instead of /data/.
• Confirm that the old /site/ still works.
• In svn (and then in eXist), rename /site/ /site_old/.
• In svn (and then in eXist), rename /redesign/ to /site/.
• Confirm that the subdomain site is now showing the redesign code.
• In controller_config.xml, set up /site_old/ so people can see the old version of the site.
• In svn (and then in eXist), copy the new /site/ to /dev/.
• In controller_config.xml, set up the new /dev/ view so that we can work on site content without breaking the live site.
• Test everything.

Added a new Home item at KMF's request. It shows up only on small-format devices, and is required because the breadcrumb trail has now been removed from half the pages on the site, so there's no obvious link to the home page near the top (the small-format rendering does not include the linked logo).

I've been trying to figure out what's causing the aboutMoeml div to disappear on these devices, but no luck so far. It's very odd. There seems to be nothing in the CSS that would hide it.

1. If you have JavaScript turned off, then mousing over the buttons will show the menus, and clicking on them will work as expected, except that in the case of the Credits menu, when you click on a name, you'll be taken to the page for that person (rather than seeing a popup box, which is dependent on JavaScript).
2. If you have JS turned on (the normal state), mousing over the button will do nothing, but if you click on it, you'll see the menu. When you navigate via the page menu, the menu will disappear after every click, but for the Credits menu it will stay open until you click on the button again. Clicking on names in the Credits menu will show the popup for the person.

I've tested this on desktop Chrome and Firefox, and on Android and iPad tablets, and it all seems to work correctly.

This gives us a level of graceful-failure on the menus that is far better than that in the content itself; links in the content right now are completely dependent on JavaScript.

Started working through the still-sparse documentation for OpenLayers 3 and trying some experiments to see how practical it will be for our purposes. So far, it seems to be very closely tied to GIS coordinates; this is difficult for us, because our "map" is so far from being realistic that it's impractical to geo-rectify it, and what we really want to do is to take advantage of a slippy-tiling sort of system to display it, but drop SVG layers on top which relate to it only through pixel offsets rather than GIS.

I've created an output format which consists of virtually standalone XML, with all external references imported into the document and all links made local (except for some obvious exceptions). This validates against tei_all and is now available alongside the old "standard" XML output, in which private URI schemes are resolved to file links. The latter output is still not quite valid against london_all.odd because of the constraints on the catRef attribute values; I'll have to think what to do about that. I might make the valList suggested/open for those attributes in the schema, but further constrain them in Schematron, which would allow for the enforcement without invalidating resolved references (since the output XML doesn't link to the Schematron).

We needed a lightbox effect for showing YouTube videos such as JJ's intro to the project on the home page. After looking at the excellent CSS Ninja FutureBox, I decided to implement something even simpler based on the target pseudo-attribute, and it's now working, using pure CSS. PS has enhanced the styling and it seems to be working well and looking good. It's based on a TEI <media> element, and although I have yet to document in detail how to do the markup, it's very simple, and anyone could look at the code and re-use it.

Meeting with PS and KMF to work on home page design; implemented many of the decisions on the fly.

I've implemented the following new categories:

mdtParatext
mdtParatextNews
mdtParatextBlogPost
mdtParatextPeople
mdtParatextCv
mdtParatextLegal

Where documents had the old mdtAbout or mdtAboutProject values, I have replaced them with Paratext ones, but many of the documents in /info/ still don't have appropriate document type <catRef>s, so someone will need to go through them and make sure they all get the right values.

I'm wondering whether it would be a good idea to split /info/ into two separate folders even now, and tweak the old site to handle it. The natural split would be this:

/info/
/documentation/

There is already a "documentation" folder, but it really doesn't contain documentation; I'd suggest renaming this or even moving it. The stuff in it doesn't belong on the website, but it regularly gets uploaded there.

We had a problem that showed up in IE and Chrome, where the fancyScroll() function, which scrolls the page to an appropriate offset so that the target element is visible below the top menu, rather than hidden behind it, was failing on longer pages; the offset was not apparently being used in the scrollTo() function. I think the reason for this is something to do with the point at which page dimensions and element offsets are calculated during a page load; Firefox was accurately calculating them, but the other browsers were apparently unable to do so during a function called by the onload event. The only fix I could find that would work was to use a timeout for the scrollTo() call; that seems to allow the browsers to complete their setup, and then the scroll works correctly.

I've also fixed a bug in the left column menus, where they were not "sticking open" when clicked on. Now they stay open, and if you click again then move the mouse away, they close.

Meeting this afternoon, after which I made the following changes:

• The breadcrumb trail now only shows for primary source and encyclopedia docs.
• <gloss> has a type attribute with two values (em and critical).
• Org popups now include a list of mentions of the org in other documents.
• Page contents now work correctly even when index elements intercede between the beginning of a div and its head element.
• Site chrome is now suppressed when printing.
• <lb type="hyphenInWord"> now renders a hyphen.
• Lists of livery companies which are only composed of orgs with @type="greater" now sort by their subtype attribute.