[0x3b0b]It feels weird to me that the conversation on the fedivers:creator MR dropped off with so little additional commentary after the remark about application/jrd+json versus application/activity+json, and I'm not sure why.
[snarfed]I know people here with standards body experience ([tantek]) prefer inline, and I believe them, after having implemented parsing and generating for both, but I don't have good crisp instincts or reasoning for why yet
[snarfed]yeah I guess specifically an explainer on, if there's been evolution over time on which is preferred and why, ideally from people with experience, maybe standards body oriented or not
[tantek][snarfed] I've implemented both, and there are definitely interesting trade-offs. Short answer is that inline markup tends to be more robust & flexible over time which makes it more usable, forgiving, and thus user-friendly (author-friendly).
[tantek]index or offset based annotation like that has some interesting use-cases like the TSS hackery I did with Rohit at the Twitter Annotations Hackfest which subsequently (AFAIK) scared them into not shipping tweet annotations for open developer use
LoqiTSS is Twitter Style Sheets or Text Style Sheets, a compressed form of CSS applicable to short bits of text, suitable for storing in very small fields like 512 bytes https://indieweb.org/TSS
[tantek]AFAIK all "longer" writeups/analysis of those approaches are pre-web, back when SGML was one of many approaches to marking up / styling things. I think there were plenty of such analyses but in the 1980s and thus impossible to web search for (since Google killed Netnews search which would likely be a good source)
[tantek][snarfed] one of the more interesting use-cases back from those days was tables, as an alternative to CSV/TSV, having a side-file indicate which character offset numbers on a line "meant" which columns
[tantek]since keeping the monospaced tables looking visually correct reinforced preserving the column offsets, that particular use-case was more longterm reliable
[tantek]and you could "load" the whole thing into a 2d array datastructure (or a 1d array of strings) and quickly do substring lookup for specific fields
Loqi[preview] [Eric A. Meyer] I just saw someone characterize running a Mastodon instance as being “no harder than running your own mail server” and I thought “off, that’s a savage takedown” before grasping that they meant it as a positive.
jimwinsYeah, it's actually pretty savage. Actually getting a mail server running might be easier, but you'll quickly hit the problem of the big mail providers not wanting to talk to your server.
jimwinsfediverse:creator could be seen a signal of that sort of problem being on the horizon for Mastodon, as well as seeing what Meta/Threads does in the name of the moderation problems.
[KevinMarks]Google Wave used the offsets thing as a way to make the Operational Transforms more tractable , but that also made interop harder. As the bsky example shows, how you count offsets is hard with the variable size of what is considered a character by different programming languages, and if you get the offset wrong you can break encoding as well as layout
IWSlackGateway, [tantek], [Paul_Robert_Ll], [KevinMarks], [manton], [benatwork], [jeremycherfas], [snarfed], gRegor and [qubyte] joined the channel
[tantek]Heads-up to folks here who want to participate in W3C TPAC and could benefit from funding assistance (e.g. if you don't have a company paying for you to go), applications for the Inclusion Fund for TPAC 2024 are now open: https://gzobeteisdd.typeform.com/to/PS8Jz5Rj
[tantek] [benatwork] do your posts still accept webmention comments? E.g. on https://werd.io/2024/substack-rival-ghost-federates-its-first-newsletter-techcrunch ?
[benatwork] Something's currently broken with webmention. Thanks for the prod - I'll take another look at fixing
Loqi TSS is Twitter Style Sheets or Text Style Sheets, a compressed form of CSS applicable to short bits of text, suitable for storing in very small fields like 512 bytes https://indieweb.org/TSS
jimwins Bluesky calls them "RichText facets" and Paul Frazee has a blog post explaining why they chose it: https://www.pfrazee.com/blog/why-facets
[KevinMarks] Google Wave used the offsets thing as a way to make the Operational Transforms more tractable , but that also made interop harder. As the bsky example shows, how you count offsets is hard with the variable size of what is considered a character by different programming languages, and if you get the offset wrong you can break encoding as well as layout