Ben Jonson 2025-12-01 to 2025-12-05
to : Martin Holmes
Minutes: 610
On Monday, worked on putting in place all the various ancillary files that were still missing, some from the crawl and some from the repository; PDFs, MIDI, and more images are now working well. Then started looking in detail at how the Works are rendered. It’s a very arcane process depending on a Cocoon-style XSLT pipeline which is going to be very hard to staticize. The TEI is rendered into HTML in response to JS calls from inside JQuery. The XSLT is in the repo, but the chain of processing will be difficult to reproduce, and it might be simpler to rewrite this section of the site if we can, making it simpler on the front-end as well as static.
On Tuesday, posted a copy of the site-so-far so that it can be examined by MB and tested by others. Then started looking at the bits that aren’t working which are not plays or poems, and got the masques up and running. Essays are going to be a bigger problem, but we have permission to nuke the paginated views, which will reduce the number of files to be processed considerably.
On Wednesday, rescued the Foot Voyage page, and started work on the Essays,
which I now have basically working, although none of the filter controls are
working anywhere. My strategy now is to combine the paginated listings into
a single page, and then I’ll take the type and related filter tokens from
the original XML files, and migrate them into the listings files as attributes
on the table cells, then add some JS to filter the listings tables based on
those attributes. It’s crude but it’s practical, and it reduces the number
of pages overall, as well as making it possible ultimately to get rid of
all of the browse folder content and migrate everything up
one level, where some of the key content already lives. We can also use these
values to create <meta> tags in the page headers for
staticSearch to use, ultimately.
On Thursday, started working on pre-deriving a map of all the categories
to documents and vice-versa; this is a little tricky, and it may be undermined
by the fact that we never actually completed the full crawl of the original site
because of the size of it, but I believe that what was not crawled was basically
the massive number of pages on which combinatorial results from multiple category
selections were listed; this means that, providing we got all the basic ones,
the combinations will follow trivially. By the end of the day, had a process
that creates the mapping document, which the later build step can use; and
got it adding document ids to the <html> elements based
on that info.
On Friday, added document ids to the HTML output, and then staticSearch
<meta> tags in the header. Then put the document categories
into a @data-cats attribute in the table row containing a
link to that document, so that listings pages can be managed using JS.
Finally, I worked on getting the performance images, which were badly
lazy-loaded by JavaScript for some reason, working using normal @src
attributes.