#dev 2023-05-25
2023-05-25 UTC
# 
[0x3b0b] I suspect there are a lot of little reasons that also contributed. One thing I think I remember reading at some point was that it was influenced by underline being uncommon to use for emphasis in electronic media, which in turn was influenced by the browser default of rendering links with an underline and people therefore tending to expect underlined text to be a link.
# 
bkil Well, I'd say the explanation is just the opposite: browsers had chosen to underline links as it was not typically used for text markup. HTML didn't include any of i, b, or u in its first draft: https://www.w3.org/History/19921103-hypertext/hypertext/WWW/MarkUp/Tags.html And while standardizing i, b, em and strong, had only mentioned underline as an optional or "possible" feature to implement https://datatracker.ietf.org/doc/html/rfc1866#section-5.7.1.3
# bkil Dumb tty screens similarly only contained a single font. They gained the ability to display colors (or shades of gray) for emphasis and underlining was typically also supported due to the way the cursor can already be rendered in hardware and how the bottom half of characters were reserved as a calm space.
# bkil bottom 1-2 lines
# [0x3b0b] Wouldn't surprise me if it perpetuated as a cycle that started in print. I'm pretty sure I remember style/design recommendations in the mid-late aughts that suggested you should avoid underlining things that weren't links on web pages because people would try to click on them and be annoyed.
# bkil That's correct. But that's way into the mid nineties.
# [0x3b0b] Let's see. I would have first started having any interest whatsoever in the creation of web content around 1997, probably in doing it well or according to any conventions or standards in late 2000 at the earliest and more likely not until 2005, until at least 2008 I would have been mostly constrained by predetermined choices, and I'd guess that I first encountered Textile in 2009. So yeah, I was late to that conversation.
# bkil Says here that ANSI escapes/VT100 did specify how italics could be marked up, but support for that was poor in comparison to underlines https://en.wikipedia.org/wiki/ANSI_escape_code#SGR_(Select_Graphic_Rendition)_parameters
# bkil Then again, we were talking about how an author could mark up content for themselves in a lightweight textual format, not how the web page should appear. I.e., using _ in the markup could appear as italics for most web users and as underline for Opera Mini users or those using a text browser.
# bkil Similarly how your markup for hyperlinks should be clickable on the web and in postscript and PDF (preferably behind a meaningful and informative phrase), while it should appear in full or as a footnote in print. You could generate both formats from the same source.
# bkil So many great ideas collected here: https://en.wikipedia.org/wiki/Italics#Substitutes
# bkil https://en.wikipedia.org/wiki/Underscore#%22Simulated%22_underlines_in_plain-text https://en.wikipedia.org/wiki/Slash_(punctuation)#Text_messaging Using asterisk for bold feels quite overloaded, though: https://en.wikipedia.org/wiki/Asterisk#Typography https://en.wikipedia.org/wiki/Emphasis_(typography)#Punctuation_marks
JanMiksovsky, KMF, greglopez, IWDiscordRelay, mouse[d], petermolnar, gRegor, gerben, IWSlackGateway, holiday_medley, geoffo, [tantek] and holiday_1 joined the channel
# 
[tantek] [snarfed] this one sounds like it's for you (and me and everyone else that uses Bridgy Fed) https://mastodon.social/@Jeremiah@alpaca.gold/110429738257155546
jonnybarnes joined the channel
# [tantek] from https://indieweb.org/Bridgy_Fed#IndieWeb_Examples kongaloosh [gRegorLove] jamievtanna aciccarello
gRegor, [jacky], [schmarty] and [snarfed] joined the channel
geoffo joined the channel
# [snarfed] on an unrelated note, "The expectations of people on the Fediverse are sometimes just weird" 💯 https://alpaca.gold/@helge@mymath.rocks/110429811997721192
# IWDiscordRelay <capjamesg#4492> FEP?
geoffo joined the channel
# [tantek] that being said, if it's a group of implementers choosing to use it as a mechanism/process to discuss things amongst themselves to interoperate, it really doesn't matter what it's called or how it works, because they are motivated to continuously improve it to achieve interop as their actual goal, not a social club
sebbu and [KevinMarks] joined the channel
# 
[KevinMarks] PEP's for Python have worked well, but ti depends a lot on community stewardship
geoffo joined the channel
# bkil aaronpk: About user subdomains vs. handles. The reason historical was probably that around the first heydays of The Fediverse around 2008-2012 where PHP-based implementations dominated, it allowed admins to easily and cheaply self-host an instance on shared (free) web hosting plans. Supporting subdomains requires more complicated implementation, quite a bit of fiddling and also only possible on a VPS.
# bkil Although this went out of fashion long ago, Friendica still supports not just users living under a subpath, but even installing the whole Friendica instance under a folder of its own instead of on our own subdomain for this reason (i.e., older and cheaper shared hosting plans did not give you even a single subdomain of your own and some tildeverse hosts still don't).
[pfefferle] and geoffo joined the channel
# [tantek] curious that none of the "direct" implementations of ActivityPub in the examples here use just domain as username: https://indieweb.org/ActivityPub#IndieWeb_Examples — I wonder why that is. aaronpk, did you document why you chose your email-like `@aaronpk@aaronparecki.com` rather than @aaronpk.com@aaronpk.com or @aaronparecki.com@aaronparecki.com ?
# [tantek] posted: https://tantek.com/2023/145/t1/
# [snarfed] sadly one minor nit there is that the user mentions end up as just links, not real AP user mentions. we're still struggling to get mention interop working, https://github.com/snarfed/bridgy-fed/issues/493
jarenado joined the channel
[manton] joined the channel
# 
[manton] This is barely IndieWeb-related, but… I’m playing more with Nostr and wondering how far I should go to securely store someone’s private key on a server. Unlike passwords and tokens which can be revoked or changed if compromised, a Nostr private key _is_ your account, so if it’s leaked your account is ruined and you have to start over with nothing. Thoughts?
# bkil The common solution to this is to password protect the key file before storage.
# bkil And also introduce a new account (with new private key, etc) on the server if it needs to relay data in a MITM position.
# bkil If your question was for me, type in ssh-keygen or gpg --gen-key. It will ask for a password for storing your key in a file.
# bkil You can store your ssh/gpg key on a hardware token that requires keying in a physical PIN after inserting into your computer as well as an alternative. What is the use case again [manton] ?
# bkil I mean, Matrix also supports backing up your client side E2EE encryption keys on your HS after encrypting it with a user password. I can search for a reference if you want, but it's common knowledge as Element asks for it on first setup.
# bkil Precisely why we have added properties to our analysis on https://bkil.gitlab.io/secuchart/ about Account recovery after device compromise and Account deactivation after device compromise
# bkil Note that solutions would exist to this in decentralized systems, it's just that existing implementations usually don't prioritize it.
# [snarfed] yup. coincidentally I was reading https://github.com/nostr-protocol/nostr/issues/45 literally moments ago. the answer from fiatjaf and much of the community seemed to boil down to, "out of scope, users should take care of their keys, git gud"
# [snarfed] back to secrets storage, [manton] in case it helps as moral support, I've had to store users' sensitive API tokens in Bridgy for 11y, and AP private keys for Bridgy Fed users for 6y, and it's fine. be security conscious, do the obvious right things, publish a responsible disclosure policy, and you'll generally be ok
# bkil Without any of these, you basically have to keep track of how you had originally made first verified contact with each peer in the past (preferably printed on paper or by heart) and repeat it (i.e., manual revocation propagation).
# bkil Odd that you mention this, I just had an idea of keeping the keys and private data of your clients decrypted for as little time as possible in memory every day (every week) in rotation to reduce the surface for exposure and attack https://github.com/bkil/freedom-fighters/blob/master/en/server/encrypted-vm-storage.md#user-content-scoped-encryption
# [tantek] thanks for the wikipedia links and the key phrase (no pun intended) [snarfed]. This seems like a good place to start with this sort of thing [manton]: https://en.wikipedia.org/wiki/Key_management
# 
capjamesg [snarfed] I just saw https://www.thoughtworks.com/en-gb/radar that you mentioned on bsky. That's cool!
# 
Loqi capjamesg: [tantek] left you a message 21 hours, 8 minutes ago: during HWC you asked how I markup (or should) the virtual headings I have for Glossary and References and I noted they are plain text. The longer answer is, they’re a part of "block level" auto-markup (in contrast to inline level auto-linking etc.), and I have some ideas for that in the latter part of the table in https://tantek.com/w/Markdown#BeforeandAfter
[chrisbergr] joined the channel
# [snarfed] whoa, new Twitter API tier just dropped, 1M tweets/mo for $5k/mo, https://twitter.com/TwitterDev/status/1661790253886177280
[snarfed]1, [jacky]1, [pfefferle]1 and bret_ joined the channel
geoffo, benji and jonnybarnes joined the channel
# [KevinMarks] [capjamesg] bsky.link is 502 again - did you push my patch?
jeremycherfas joined the channel
# 
[jacky] There's a solution that's been floated around (that I like, tbh) and have slowly been squeezing to Sele via https://darkcrystal.pw/
cdravcte joined the channel