#microformats 2023-02-08

2023-02-08 UTC
#
@wastemobile
↩️ 不到兩個月的時間內,Mastodon 月活躍用戶從38萬攀升至250多萬,又降回140萬,許多新用戶抱怨「很難用」。 是事實,此波震盪最重要的並非尋找一個推特替代品,而是如何讓 ActivityPub 互通各類 microformats 網路原生內容的環境成形。 The Mastodon Bump Is Now a Slump https://bit.ly/3RGRYeO
(twitter.com/_/status/1623123605281456128)
angelo joined the channel
#
[0x3b0b]
Right. Carrots it is.
angelo, ur5us, ur5us_, gRegor, janboddez, [dave], [pfefferle], [tantek], [tw2113_Slack_], IWSlackGateway, ehmry, [snarfed], [manton] and [KevinMarks] joined the channel
#
[tantek]
I found a use-case for keeping *some* relative links relative in e-content: footnotes
[hollie] joined the channel
#
[tantek]
e.g. an inline footnote fragment link to its note expansion elsewhere (near the bottom) of the e-content, which then may have a fragment link back to the inline reference. that would ideally still "work" as fragment links to scroll up/down into view accordingly, even if that e-content made its way into a feed reader, or say Mastodon
#
[KevinMarks]
are they fragment-only links?
#
[tantek]
correct
#
[tantek]
I mean yes
#
[KevinMarks]
that seems doable as a parsing heuristic
#
[tantek]
really funny to be solving these two different (seemingly unrelated) problems at the same and find a connection
#
[tantek]
also, fragement-only links inside an e-content can be reasonably expected to reference other things in the e-content, in general (whether footnotes or others), so keeping them fragment-only makes sense
#
[tantek]
I forgot the issue number
#
[tantek]
will update the proposed resolution
#
tantek
edited /microformats2-parsing (+103) "editorial: s/whitespace/spaces, and add note in HTML parsing rules clarifying existing meaning of 'spaces' meaning HTML collapsible blank spaces (which is what’s implemented)"
(view diff)
#
tantek
edited /microformats2-parsing (+195) "Proposed for e- parsing, normalized absolute URLs in all URL attributes except those with fragment only values, to solve issue 38 and allow for relative links to *inside* e-* properties to keep working in embedded use-cases"
(view diff)
#
[tantek]
^ [snarfed] gRegor Zegnat proposed way of resolving https://github.com/microformats/microformats2-parsing/issues/38 while preserving a very useful e-* property embedding feature (relative links within the e-* value itself), use-case: https://indieweb.org/footnotes 🙂
#
Loqi
[preview] [willnorris] #38 resolve relative URLs in e-* html
[jamietanna] joined the channel
#
[tantek]
[will] also
#
gRegor
Hm, what if the reader has 2+ posts with the same fragment? You were talking about making yours something like #note-{number} right? so they would repeat among different posts.
#
[tantek]
no you missed the detail about a prefix
#
[tantek]
search for "concatenating prefix param" here https://indieweb.org/footnote
#
[tantek]
each of my posts will have unique prefix params to solve that problem, probably based on the newbase60 compression of the date and the post number
#
gRegor
From the parser impl perspective, it seems more difficult to only make some links absolute. Not a lot more difficult, but some amount.
#
[tantek]
a little yeah you have to add an if check for the attribute starting with '#'
#
[tantek]
and not just a reader having 2+ posts, this needs to work on my home page 🙂
#
gRegor
Not sure, my initial thinking is I would expect clicking a link in a reader to take me to another page, not jump within the reader itself.
#
gRegor
And same for masto
#
[KevinMarks]
it is a bit odd when the footnotes are actually links, yes
#
gRegor
But really I've only experienced reading your posts in a reader
#
[tantek]
I think it's distracting if trying to view a footnote changes your reading context
#
[tantek]
this is kinda new territory as I don't think many (any?) blogs have supported this sort of behavior in readers etc.
#
[tantek]
so some amount of expectation changes can be expected. I like that the small unicode number hints at, this is a very small link and may just do something a little different than clicking a whole big hyperlink URL or word/phrase which will navigate you
#
[snarfed]
so the tradeoff seems to be, pro: keeping fragment links within the current embedding is nicer than loading the separate original page
#
[snarfed]
con: that experience is going to be brittle and inconsistent in practice, due to sanitization varying widely, fragment destinations hidden in collapsed longer posts, etc
#
[snarfed]
given that tradeoff, I think I lightly lean toward fully qualifying fragment links
#
[snarfed]
but only lightly
#
[KevinMarks]
then you end up with the weird mastodon thing of relative links to the cached full link on the original instance
#
[KevinMarks]
not sure how you unwrap that, but it is probably what is breaking the inline ones
ur5us_ and [schmarty] joined the channel
#
[tantek]
hence per #indieweb-dev the opportunity for experimenting and learning!
[aciccarello], [asuh] and barnaby joined the channel