KartikPrabhu_Bret you don't need an annotation system to use marginalia.js make a comment form that takes a reference phrase as input and use the corresponding fragmetion as a data attribute and marginalia.js does stuff for you
gr0k, reedstrm, npdoty, fmarier, el3ctron, case1, RichardLitt, xxcoeurxx, lukebrooker, aaronpk_, jdp23 and chrissaad joined the channel
reedstrm@sigh@ so tilde.club is "retro" web, basically nostalgia for "simpler times" copying the old look and tools, though w/ a non-corporate, community feel, while indieweb is reinvigorating the principles of the old web, not the look or technologies.
reedstrmgRegor` no argument from me on if it's fun, and retro and "cool" - just, it's not "own your own data" just cause you used 'vi' to edit it. On someone else's server.
kylewm"When the server went away (due to some poor planning and communication on my part) I lost the app, the database, and - even more frustratingly - the database backup."
danlykeReposted from #indiechat: Further thought on yesterday's IndieWebCamp wiki logins and multiple identities per indie-web domain thread: Just realized another use case (beyond married couples with a pre-nup concerning distribution of virtual assets) for multiple identities: I'm "Facebook friends" with a number of my friends' pets...
danlykeThat seems to be a need for separate identities that isn't as separate as our online web identities sometimes are (ie: I'm a different "Dan Lyke" on LinkedIn than I am on Facebook, but pretty liberal about those two tying back together, I'm an entirely different name on MetaFilter and Sensible Erection...)
kylewmdanlyke: I think those two identities would be two different domain names, right? you wouldn't want to be me@danlyke.com for one and anonymous@danlyke.com for the other
gRegor`danlyke: I haven't seen the conversation you're referencing, but you mean support for example.com/user1 and example.com/user2 when using indieauth?
danlykekylewm: Exactly, those two identities for me are fairly diverse, but I would totally want www.flutterby.net/User:DanLyke and www.flutterby.net/User:MyCat (if I still had a cat)
danlykegRegor` yes, this came up because my OpenID/IndieAuth/etc identity/login URL is http://www.flutterby.net/User:DanLyke in anticipation that my wife's will be http://www.flutterby.net/User:CharleneMarie when she needs one. Discussion yesterday was about how support for multiple users per domain was kind of an edge case, and several such domains had been blocked from indiewebcamp.com login.
danlykeAh: the "DNS cartel-oriented thinking" is the "Tie identity to an identifier that we lease from the DNS cartels" problem. Really, it's not a whole lot better than a example.com/~username identifier that we lease from an ISP.
iccoQuestion: How are people dealing with posting content from their site to social networks? Manually? Synchronously? Or some nice asynchronous setup?
danlykeiccio: I insert into a table for "statusupdates", and then record numbers from that table into "update_facebook", "update_twitter", "update_flutterby" etc., along with a timestamp and a delay factor. I then have a queue runner that runs every 5 minutes and attempts to push those various updates out, deleting the records from the update_* table as those posts succeed.
danlykeI used to, before Twitter started auto-shortening. I do a little bit of prep work, if it's longer than 140 characters and the Twitter shortener wouldn't make it shorter, I crop at the first space before 118 (or the https?://) and insert a link to the full entry on my site.
danlykeI also explicitly extract links and send them to Facebook as a "FEEDLINK" rather than a "STATUS", so Facebook grabs the preview (makes posting pictures to Facebook actually work nicer for the end user than the native Facebook UI).
danlykeI use the PHP utility "fbcmd" to do the Facebook posting. A bit of a PITA because Facebook is getting away from authorizing non web apps, so I have to re-auth regularly.
snarfed1danlyke: agreed that fb's new(ish) 2mo token expiration is annoying for non-web apps. for bridgy, i actually ended up generating fb notifications so users know they need to re-auth. https://github.com/snarfed/bridgy/issues/59
danlykeOh, and I use the separate tables for the update with the premature optimization notion that an index scan on a boolean value was probably less efficient than a separate table with "this needs attention".
danlykesnarfed: Good idea. I've just been waiting 'til my posts don't show up, but I should set up a tool so my wife can easily re-auth her side of things... and this is how command-line tools and quick hacks become unmaintainable nightmares of interconnected little scripts strung across multiple servers.
snarfeddanlyke: heh, yeah. also, usability and maintainability req'ts are much different if it's just you vs other users. for just yourself, something as minimal as a recurring calendar reminder may be enough as a reminder to re-auth
danlykesnarfed, I have webmention meta tags for inbound, and I have some half-finished code for outbounds, but I'm still skeptical on the whole notion of webmentions (ie: Referer tracking v.3).
snarfeddanlyke: sure. i'm not religious about webmentions specifically, they just happen to be the trigger for bridgy publish, which warns you when its auth expires, in case you're interested in trying it
danlykeYeah, meta-refresh seems like a bad idea. I've played with some of the real-time Node frameworks, but... Should probably learn those rather than hacking something together with http://meteorserver.org/
danlykeben_thatmustbeme of course the other thing I've thought about is: Why are we using HTTP for our sharing, rather than XMPP or RetroShare's protocol or BitTorrent Chat or something built for real-timeness.
danlykeYep. All the Flutterby.com posts went out to NNTP, and comments came back. So, to your point, all the permalinks that still exist are HTTP (if I still had the server up, they'd also be available via Message-ID)
danlykeBecause we're having this conversation over a channel that's not HTTP. I'd prefer it over email or NNTP, because both of those have local caches and permanent archives and are more easily searchable and organizable.
danlykeAnd, because as of right now: There are about as many people in my social circles that I'd like to engage in conversation running any of these other hypothetical protocols as there are running WebMention.
danlykeAnd a hell of a lot more of those folks are running RSS and Atom than microformats (which is why I'm scanning for inbound links from those feeds and haven't yet bothered to put microformats in my parser)
danlykebut RSS, Atom and Microformats all extend HTML with additional semantic content. It's just that the former are carriers for HTML (and that's a "ugh" decision, but water under the dam), and the latter is the carrier.
danlykeIn a perfect world, I like Microformats better, but from a practical standpoint I need more infrastructure before I can take advantage of the advantages of Microformats over RSS and Atom.
danlykethe first application I've got is outbound (and why I've implemented it in my blog main page, and in the HTML feed of my personal status updates). That's a matter of putting an RSVP tag in some entries and sending Webmention pings.
danlykeUnmung goes the wrong way, though. I already have the infrastructure for RSS/Atom. That's easy and well supported. I need to write the Microformats parser yet.
danlykeSo how do I do the webactions <indie-action ...> tag thing without silos? I don't care if they can star it on Twitter, I'll see it there. I want them to be able to star/reply on their own site and put it under mine.
danlykeTantek, you'd mentioned a mime-type or something similar that the browser could use to redirect the click to the user's site (and maybe default to my own local reply/like mechanism), what was that?
danlykeBecause the existence of this capability on other people's site is a reason for me to get off my butt and get MF parsing integrated into my infrastructure.
danlykegRegor` yeah, me too. In fact given my experience with Referer/Trackback/Pingback, I'm guessing it's easier to manage spam with local blog comments. But Tantek is optimistic, so I'm willing to try...
danlykegRegor` yeah, in practice just not having Wordpress input names has been enough that I've been able to manage the spam with the occasional DELETE FROM weblogcomments WHERE id=...;
danlykeWebmention thus far is more obscure than my blog. I'm guessing every URL on my site gets several Pingback/Trackback spams a day, just as POSTs, without any discovery info in the HTML (I disabled those endpoints a decade ago)
gRegor`I don't know if I'd say that's the main use case of webmention. It's a popular one, because there's so many more people on silos than have their own sites with webmention support. It's nice that I can respond to kylewm directly on his site though instead of going through twitter.
gRegor`Re-phrased: I'm personally more excited about being able to comment on other sites via webmention than I am pulling in backfeed from Twitter. I'll see the Twitter stuff anyway, and honestly who favorited a note is not going to be very important to me in 5 years. :)
danlykekylewm For cross-site scripting protection purposes you should be putting a nonce in your forms anyway (confession: I haven't yet), so there's no reason to just generate nonsence form names for every page view.
danlykegRegor` no idea on WordPress nonces, I'm just assuming that the field names my spam logger is catching are Wordpress names. (I log for a bit, find the major offenders, "ufw deny" them, traffic on the server drops dramatically)
danlykeSo for real-time chat, we would need: A way for a user to discover that they can comment on a status update on my site. Them typing up a response and sending a Webmention to my site. My site getting that webmention, grabbing the appropriate h-whatever section of their HTML page, and incorporating it into the response stream, right?
danlykeCurrently 1 (discovery) is "some text in the footer". 2, generating the response is up to them. The webmention/h-stuff is my responsibility. Seems like (barring my concerns about spam), we're kind of stuck at #1.
danlyke(I know a few people online who might, but they don't update their blogs any more and just live on Twitter and Facebook, so I may be overestimating their dedication to the cause)
gRegor`For those who are on silos, if you're using /POSSE and /backfeed, they need no instructions at all. They just interact with the silo copy of your information.
danlykeI guess I'm still skeptical about "people with a greasemonkey script can comment, everyone else gets sent to a silo". Maybe I need to implement local comments so that those examples don't seem like Silo traffic drivers.
gRegor`tantek had suggested to me at one point sending people to Twitter to reply instead of local blog comments. I didn't care for that and the local blog comments are still worth it to me over the small nuisance of spam, so I agree I'd rather not send people to a silo
danlykekylewm hooooookay :-) Guess I need to set up my Webmention thingie to do the "excerpt the remote content" bits, and when I'm reasonably certain I'm there ask one of y'all to test it.
danlykeStill don't have a good excerpter, probably need to harass the people who's RSS feeds I check for inbound links to put microformats in their pages so I can grab summary text rather than titles.
danlyke(Interestingly, I got it right for attribute names, and I allowed ":" for namespacing XML in that portion of the cleaner. Now I have this awful desire to go read specs and find out which HTML or XML version didn't allow '-' in element names.
danlykeKevinMarks__ yeah, my backlinks structure doesn't have a body part yet. The semantics for what I should extract from the remote page for that are what I need to add...
danlykeOh ugh. I need to do HTML sanitization on that body, don't I? Okay, pretty sure I've got code that does that, but also pretty sure I haven't looked at it in close to a decade...