The site now has polygot markup and can handle both XHTML5 and HTML5
parsing rules. My staging site will be XHTML but my main site will be
HTML5, just in case of parse errors.
If other tools (e.g. LightHouse) end up supporting XHTML5, I'll consider
switching the content-type to XHTML.
I go out of my way to make sure the site works on viewports well under
240 CSS pixels wide. Some pages are very long, so a single mistake in
one part of the page shouldn't cause small text throughout the page.
Some article extractors get confused when the article body references
another author with microdata. Declare the author itemprop a second time
to make them happy.
Add an RSS feed for notes. Next up, replacing the RSS navlink with a
page containing links to both my posts and notes RSS feeds. When I get
ATOM and WebSub, it'll have links ot those too.
Also fixed some typos and switched "Posted" to "Noted" in the context of
notes.
- Stop using draft WAI-ARIA 1.3 that isn't supported yet
- Make in-page links focusable across shortcodes/partials
- Replace existing in-page heading anchor links with a more accessible
option.
- Make backlinks aria-labelledby instead of giving them an aria-label,
so they can be translated.
They're not interactive elements and there's no hint that they should be
hovered on. Wrapping them in an <abbr> would just make the markup even
more overcomplicated, so I just dropped the titles.
Allow specifying open graph images for individual pages. Change default
site-wide open graph image alt-text to an empty string since the default
image only has visual value, and is better off hidden from assistive
technologies.
Set an open graph image for two articles.
- Streamline CSS to reduce duplication
- Better backlink accessible names for endnotes with multiple backlinks
This required updating a false positive filter in my vnu jq filter.
- Change a quote of a code snippet to a code snippet with a citation
- Mention tkhtml
- Mention pandoc and printfriendly as exmples of moving figure elements
like floating blocks.
- Make TPGi a publisher, not an author.
- Use brackets to clarify a reference in a quotation.
- Add personal example of why it's important to test both low- and
high-end displays.
- Identify the transcript with a subheading, ending in a period so that
the aria-description separates the identifier from the concept.
- Put the transcript in a <section>
- Snuck in a trivial syntax fix for the generator element.
- Make webring links touch-friendly and accessible by using spaced-out
details elements.
- Make details elements touch-friendly by indicating interactive region
area and making summary padded.
- Sort featured posts by featured order.
- Ensure that at least one non-interactive tappable region exists on the
screen at all times, 48x48 px.
Increased font size to decrease chars-per-line (SC 1.4.8) and increase
tap target size.
Pad the nav links more and add some extra space between them to meet SC
2.5.5.
Split up some verbose and repetitive templates to use nested templates.
Cache some templates that don't need to be re-built every time.
Reading time: switch "a" to "an" when followed by the number "8".
Improve whitespace alignment a bit.
- Better print stylesheet, now with a file dedicated just for print
style improvements.
- Hide extra stuff in print.
- Bring back navbar for print because it also tells users the current
section and the site name.
- Dark theme: make superscripts bold instead of using a higher-contrast
color.
- Don't auto-minify HTML but minify other resources, inc. the RSS feed
- Manually optimize whitespace a little to balance readability and
compressed size
- Goldmark 1.4.12 switches footnotes from a <section> to a <div>; update
regexes and stylesheet to account for this.
- Goldmark 1.4.12 allows multiple footnotes with the same reference; use
that.
- Clean up templates for unminified output. Also delete an unused class.
- Switch to unminified output by default.
- microdata for CompleteDataFeed on /posts.html instead of DataFeed
- make the home link <strong> when it's the current page, just like the
other navlinks.
Create a for <figcaption> and a render hook for code snippets inside
figures, replacing some ugly and complex inline HTML in my markdown
sources.
The only visible change is slightly worse HTML alignment and programming
language indicators (with microdata).
This removed the need to use one of the regex replacements in the
processed_content.html shortcode, and increased the minimum required
Hugo version to 0.93.
WCAG recommends telling visitors about their current place in a site's
hierarchy. All pages are exactly zero or one level below a section, so
simply emphasizing a member of the navigation links should be
sufficient.
All pages except bookmarks can now fit on your smartwatch without any
adjustments, so we can disable the auto-zoom-out-to-match-mobile
wizardry that watch browsers do (comparable to the
zoom-out-to-match-desktop stuff that early mobile browsers did).
Yay.
- Shrink some excessive alt text
- Remove some redundant links
- Screenreaders that split elements up aren't just on touchscreens
- Mention ChromeVox in list of screen readers
- Move TOC higher in page
- Spelling
- Drop more unused classes
Putting the heading in the navigation element makes the structure more
logical and matches the behavior of most other websites.
Move it before the introduction in my web best practices post.
Order is significant for the ToC and post list so make them ordered.
I opted to make post-lists a reversed list, so I don't end up having
every post change its number every time I post.
Visitors can see that on post update dates; I can see that in
builds.sr.ht.
This results in not having to upload every page on a trivial change. Add
the "-c" rsync flag to take advantage of that. Doing so should also
result in longer-lived cached copies, with etags and last-modified
headers.
- Even less halation for dark theme
- More contrast for borders
- Slightly larger font, fixes APCA contrast issue for <small>
- Make responsive navbar work in NetSurf
- Make aria-current page bold
- Use content-visibility to unload footers and endnotes
- Add aria-labels to unclear webring link text
- Replace <hr> elements with css borders; the semantic meaning of <hr>
was unnecessary with section breaks.
- Some webrings I might eventually join require sending the origin via
referrer. Make per-link exceptions to my "no-referrer" policy sent via
HTTP header.
- Add hotline webring
- Add criteria for me to join a webring.
No browser I know of falls back to a PNG when given an SVG with a PNG
fallback. The Tor version of my site uses no SVGs.
Exclude the SVG favicon in the Tor version of the site, since the Tor
Browser disables SVG on the safest setting.
- SearchMySite.net uses the "keywords" meta property to categorize
sites, so include keywords on the homepage
- Exclude the published date if it doesn't exist.
- apple-touch-icon not needed, will be fetched from doc root
- re-order <head> elements to optimize for compression algorithm size
savings. shaved off like 15 bytes. this was a good use of my time.
- Remove reference to unused syntax.css
- Stop Apple's magic phone-number-linkification. If I need to link a
telephone number I'll use a tel: URI, thank you very much.
The favicon is 175 bytes, smaller than the size of the HTTP headers to
fetch it. It can be inlined.
Now, pages that don't have any other images need just a single request.
The main stylesheet for the sit is just 721 bytes uncompressed. I can
inline it safely to shave off a request, since the headers alone are
probably comparable to the size of the CSS.
- The default time is browser-default colors, not "light". The only
"theme" for my site is the dark theme.
- The "color-scheme" doesn't have wide support
- The "color-scheme" property doesn't really do anything much if you use
browser-defaults or a dark theme with @prefers-color-scheme
- Footer should contain date last built so people don't start scratching
their heads wondering why webmentions aren't showing up
- Add rel="nofollow ugc" to webmention links.
Statically grab and include webmentions during Hugo builds, no JS
involved. Hugo supports making web requests and parsing the resulting
JSON, so there was no need to use an external program either.
