Map Of London

The pageContents menu is not sticking open when clicked in Chrome.

The pageContents menu will not disappear when focus is lost on touch devices.

Did some tidying up after changes while travelling, and then worked on implementing the three new boxes for the home page. They're now basically working. The first is an "interesting snippet" box, which searches the db for <seg type="interestingSnippet"> elements and inserts one at random (there are two in METR1 for testing now). The schema has been updated for this use of <seg>, and previous usages (odd experiments, I think) have been removed.

The second takes the latest blog post and links its title, then displays its first image below, grayscaled and chopped. The third is a single-tweet twitter timeline, showing the latest tweet.

These are now ready for styling by PS.

We've been working on the still-unsatisfactory area of the taxonomy relating to "project information", and we've had a couple of false starts and half-hearted hacks that we've had to retreat from; today we decided to put together a formal plan we can all agree on and implement it in a methodical way, minimizing the disruption to site navigation and functionality. Both TL and I have submitted some ideas to JJ and KMF, and they'll consider them.

Lots of useful progress today on the "public XML", which involves enhancing and expanding the XML source to make single files more conventional (replacing private URIs with proper links, etc.). This is near to complete now, although there is of course another complete level at which this could be done: all project-internal links to database files and taxonomies could be replaced by file-internal links to content imported directly into the file.

A final phase of this would involve transforming the schema so that it doesn't try to enforce the use of particular values for e.g. attributes on catRef.

Also did some image resizing and renaming to make image-heavy pages more effective, as well as tweaking the blog post list output from mdtList.xql so that a grayscaled version of the first image in the post appears along with the teaser. The grayscaling is done with CSS -- a bit cutting-edge and slightly hacky, but it works.

Fixed dozens of instances of ref elements being used to link to people where name elements should have been used. Also did some tweaks to the mdtList.xql library so that there's a new column in the document readout showing the status of the document (whether it's been written yet or we are soliciting a contribution for it). Finally, started trying to rewrite one of the consistency check routines to make it spit out more detailed information on where bad links are found, but that's going to be a bit tricky to do without making it even slower than it already is.

We now have working Facebook and Twitter plugins -- each has their own page embedding large activity feeds, and the Twitter one also appears down the left side of the News page.

I had to do quite a lot of hacking to make these injected blocks of content behave themselves. They tend to have position: absolute set everywhere, so you have to override that otherwise the scrolling doesn't work properly; they scroll above our top banner rather than behind it. My Firefox sensibly does not display any of this content (it's created by injecting JS from Facebook and Twitter, which executes in the context of our pages, of course, so I would expect my security settings to prevent it). Other people's Firefoxes seem to show it, though, and it works on my other browsers.

Also fixed a couple of other things such as image sizing and schema issues for TL.

Worked out many details regarding menu structure and document linking, and implemented most of them on the fly, following the meeting. Remaining to be done:

• Find a JavaScript widget which will pull content from Twitter and sanitize it for display in a page.
• Check whether colour and grayscale filters can be applied to e.g. photographs through CSS.
• Figure out whether Born Digital really should appear in the breadcrumb trail.
• Figure out a way to tag intriguing snippets, which can be harvested and linked to from the home page. Implement Schematron to constrain the length of them.

Worked on the display of e.g. mdtUndergraduate and mdtGraduate categories in the left column. Had to edit a few files manually to get some good test cases before I started. I think it's now working as planned.

A long and productive day:

• The mdtList output for blog posts now looks like a blog post listing, with teasers and a special rss feed.
• The breadcrumb trail is now pretty much working as we hoped, with mouseover title attributes including the whole catDesc from the category.
• There's a new mdtPrimarySourceLibrary category which encapsulates all the non-Stow stuff. Many other minor changes have been made to the document type taxonomy.
• The A-Z index now includes a better rendering of the document types.
• I noticed that where documents were created from XIncludes, the respStmts for the source documents were of course not included in the credits, so after much wrestling I found a solution: the makeCredits function now parses out XIncludes in the file and follows them to retrieve their respStmts alongside the ones in the containing document. This results in duplicates, but it's better than not including them. If the duplicates are unacceptable I can dedupe, but that's going to be a bit time-consuming.

I've rewritten the breadcrumb trail code and the mdtList.xq page generator, so that:

• The hierarchy of categories listed in the breadcrumb trail is no longer painfully reconstructed from the terminal category, but read from the category hierarchy itself.
• The breadcrumb captions are now taken from the category @n attributes (so they're much longer and more descriptive).
• If there is only a single public document in the terminal category, the category actually links to that document.
• In a category listing, if there's only a single document in a category, and no subcategories, that document is directly included as a link, in place of a link to its category.
• All subcategory listings end with a link enabling you to see all documents in a category.

I think that now, assuming all catRefs and teiHeaders are correctly completed, the breadcrumb trail and category/document listings should make sense.