thegreekgeek, pavel, geoffo, claudinec, thegreekgeek_, ChrisFu and mretka joined the channel
#[0x3b0b]Further anecdotal information for the thing [tantek] and [snarfed] were dealing with: After what I presume were some edits and re-federating, Tantek's profile on my instance now shows posts that were missing before, and MBP does _not_ bubble the updates to the top
[qubyte], GWG, thegreekgeek, sp1ff, thegreekgeek_, [KevinMarks] and ttybitnik joined the channel
#pcarrierany technical reason why one would want to avoid creating feed entries with dates in the future? I'm trying to post "daily" and running ahead, wondering if I might break stuff doing this
[schmarty] joined the channel
#[schmarty]If a feed reader shows posts chronologically by post date they'll likely float atop others until their publish time passes.
#aaronpkand because of that, defensive coding of feed readers suggests they might rewrite the date to the date the post was discovered in the feed instead, defeating your entire goal
#pcarrierFWIW that wasn't my entire goal, but rather to keep 1 post per day and not have to do anything for the next 2 days
#aaronpkwell if you don't mind people reading all 3 posts on the same day then i guess it's fine
#sebbusome blogs supports posting a post at a specific date
#sebbu(and i can see a perfectly valid example of future-dated post : talking about / organizing an event (even if you could or should use a calendar for that), including when it opens, close, the map, the different events inside the events, where they are (on the map), with who (guests), if you need to register beforehand, etc...
#pcarriersebbu: yeah it's different, the post only appears on the site and feeds at that date then
#pcarrierOr rather, I would guess? Anyway not so with my static site generator, it's there on next publication. I could filter and automatically deploy every hour or something, but I won't.
#sebbuyeah, i remember using webcron for something similar around a decade ago
#Loqischeduling in the context of the indieweb, refers to the feature of setting a specific time for a post to be published in the future https://indieweb.org/scheduled
#[tantek]cc: [snarfed] [manton] [KevinMarks] and whoever else is interested
ChrisFu, barnaby and srijan joined the channel
#jimw[tantek] I'm not sure I see how noticing new URLs are redirects of existing ones would even work for a feed reader. If it sees `https://example.com/original` and then `https://example.com/new` it has no way to know if `/original` is redirecting to new unless it's hitting every old feed URL to notice. And since it's a feed reader, it's probably not
#jimweven loading `/new` until the user clicks through to it somehow. That's why you need a stable identifier like `atom:id` if the URLs aren't stable.
#aaronpkmy webmention receiving code does something similar, but it relies on the webmention sender to send an update from the *old* URL
#aaronpkotherwise yeah it requires polling all previously received URLs forever which isn't scalable
#aaronpkif the new post has a reference to the old URL, then I can use that as a trigger to check the old URL for confirmation
ChrisFu joined the channel
#jimwYeah, even just pulling new URLs found in the feed could be pretty rough if most feeders did that because it would multiply a poll of your feed into multiple requests for full pages whenever there was a new post.
#jimwNot quite the thundering herd of Mastodon link preview requests because the polls are more staggered, but still not how most feed readers work, I think.
AramZS joined the channel
#[tantek]pretty sure rel=alternate can solve this?
#aaronpkthat doesn't sound like quite the right semantics
#[tantek]as in when a feed consumer sees a rel=alternate link on a new item, if the feed consumer already "knows about" the alternate link (as in, has seen it before), then it can do a HEAD retrieval of the alternate link to see if it returns a redirect to the new item
#aaronpkvs generating the whole thing and just having the http server return only the HTTP header
#jimwYou're also assuming that the feed consumer is loading the page for each entry, unless you mean the rel=alternate should be in the feed entry somehow.
#[tantek]yes feed entries should have rel=alternate links anyway, e.g. to the HTML permalink
#[tantek]jimw, no, the Atom entry can have a link rel=alternate
#jimwSo on tantek.com, you should have added rel="alternate" links to the h-entry with the old slug, and a feed reader should have seen that, checked that it redirected to the u-url for the h-entry, and thus known the h-entry with the new u-url was an update to the old h-entry it had seen earlier?
#[tantek]jimw, no restating: "the Atom entry can have a link rel=alternate"
#[tantek]only Atom consumers have this problem, therefore only need to put fixes in the Atom
#jimwwhy isn't it a problem for HTML consumers? if my feed reader is using h-feed, presumably i'm going to get duplicate entries if the u-url for an h-entry has changed between retrievals.
#[snarfed]but if it does often on your site, that's probably what u-uid is for
#[tantek]jimw, because HTML consumers load an h-feed page and see only the current entries. problem solved
#[tantek]so far this has only been a reported problem in Atom / other feed file consumers
#jimwok, so what i'd call traditional feed readers that accumulate their own database of entries from feeds don't consume h-feed, so it's not a problem.
#[tantek]so far they don't seem to have the duplication problems of trad feed readers
#[schmarty]aaronpk's aperture consumes h-feed and accumulates its own database of entries from those feeds. it should suffer from this same duplication issue i think.
#[tantek]I suspect aaronpk would say something if it did
#Loqijf2 is a W3C Note and a JSON Post Serialization Format of microformats2 for that is optimized for h-entry consuming code, as compared to the standard microformats JSON representation https://indieweb.org/jf2
ChrisFu, [morganm], ttybitnik and barnaby joined the channel
#gRegorJust tried XRay HTML input with a different u-url and u-uid property, it only returned the u-url. So I think Aperture would have the duplication issue
#[tantek]Interesting hypothesis. I'll wait for the Monocle screenshot though
#aaronpkgot an example of a URL that has this problem? i want to check the database
#aaronpki don't know which of tantek's recent posts has had the URL change
#Loqi[preview] [Tantek รelik] Twenty years ago this past February, Kevin Marks and I introduced #microformats in a conference presentation.
Full post: https://tantek.com/2024/044/t1/twenty-years-microformats
Aside: This is an even shorter summary of that post from ~200 days ago...
#gRegorAppears twice. The full post appears only once in the screenshot.
#gRegorIn Monocle I'm only seeing 1 copy of each, -summary and -shorter2
#gRegorSo it's looking correct at a glance in Monocle
#aaronpkwhich is the same as falling off the end of the feed
#[0x3b0b][tantek] re: earlier question - sorry, I used MBP as an abbreviation for "microblog.pub"
#[tantek]does aperture parse u-uid to de-dupe things "not currently in feed"
#[tantek]0x3b0b a-ha! in my abbr cache MBP = MacBook Pro
#[0x3b0b]That makes sense. I have had a similar problem before with D2 = Diablo II, Destiny 2, D2: The Mighty Ducks...
#aaronpki'm confused, from everything i can see in the database, i should see two copies in monocle, but I don't
#[0x3b0b]There's a small group that's set up a github org recently to try to take over further development of a fork of microblogpub; one of them has a rudimentary version of post editing working. I'm looking forward to trying to merge in some of the accumulate changes and then making pull requests for them to pull in some of mine. But I digress (as is my wont.
#[mattl]2SC2 = something called Starcraft 2, apparently.. it's always Short Circuit 2 to me
#[KevinMarks]If you are migrating urls and use 301 that's an explicit instruction to remember the new url in future. If you use 302 it means remember the old one and recheck.
#[tantek]aaronpk, I suspect Monocle is smart enough to collapse per redirects
#[tantek]when it tries to get the h-entry for a URL, and the URL redirects
#aaronpkI don't remember doing that but it has been a long time since i wrote that code
#[tantek]so it's fine that the database has a bunch of crufty data. I mean databases often do
#[tantek]this isn't the first time I've changed slugs so I suspect you fixed this previously ๐
#aaronpkthis is also weird because i see code for handing u-uid but xray doesn't return that property
#jimwfinally implemented a Webmention receiver on my site. all it does is email the mention to me, but that's better than the old stub that just threw them away.
#jimwopen source, technically, but not quite to the point where anyone else could use it without having to hunt down references with my name. https://github.com/jimwins/talapoin
#Loqi[preview] [jimwins] talapoin: A small blogging system, used on trainedmonkey.com
[0x3b0b] joined the channel
#[tantek]I am thinking of adjusting my auto_link function to treat @-names as *local* unless a default @-name prefix URL is specified
#[tantek]that is linking something something like @example to href="/@example"
#[tantek]which allows the local site to treat "/@" as a nicknames cache lookup
#[tantek]this would allow me to repair and redirect any past plain @-names in my past posts that used to mean people's Twitter handles back in the day, but may be deleted or sold as of today, to that person's personal site or other preferred @-contact
#[tantek]hmm, I wonder if we could do this for the wiki as well, even if it's a bit manual
#[tantek]lol turns out I already did this for myself just over 10 years ago ๐ /@t/[tantek] let's see if those link properly in the logs