#dev 2020-11-27

2020-11-27 UTC
geoffo and [Raphael_Luckom] joined the channel
#
[Raphael_Luckom] So this is something I never noticed before about feeds,and I want to make sure I understand it right. Atom and RSS are XML formats, and they _only_ specify the feed description document--on my site it's the one called `index.xml`. Specifically, Atom and RSS do not specify any particular format for the post itself--there's no RSS/Atom "microformat" that describes the content within a page the feed points to. Is that right?
#
@JamieTanna Well, tonight at #HomebrewWebsiteClub Nottingham I managed to get my staging site using my new #indieauth server, a… https://twitter.com/i/web/status/1332087797335855104 (twitter.com/_/status/1332087797335855104)
#
[KevinMarks] Not quite sure what you mean [Raphael_Luckom] - both rss and atom let you embed html in the posts in various ways from shonky escaping to CDATA to using xhtml inline (I think Tantek is the only one who does the latter)
#
GWG aaronpk: No picture?
#
aaronpk hold please
#
[Raphael_Luckom] sorry, I don't think I'm using the right words. I see an `item` in my index.xml. It loos like this:
#
[Raphael_Luckom] <link>https://www.raphaelluckom.com/posts/mid_november_check_in.html</link>
#
[Raphael_Luckom] <pubDate>Wed, 18 Nov 2020 14:09:34 -0500</pubDate>
#
[Raphael_Luckom] ``` <item>
#
[Raphael_Luckom] <title>Mid-November Check-In</title>
#
[Raphael_Luckom]
#
[Raphael_Luckom] <guid>https://www.raphaelluckom.com/posts/mid_november_check_in.html</guid>
#
[Raphael_Luckom] <description>...</description>
#
[Raphael_Luckom] </item>```
#
[Raphael_Luckom] The `link` in that post goes to https://www.raphaelluckom.com/posts/mid_november_check_in.html . On _that_ document, there does not appear to be any atom / RSS specific markup--so the feed reader consumes the XML format, but once it gets to the _linked_ page, it's on its own and can't rely on any special markup to tell it what's what.
#
aaronpk i mean yeah rss doesn't really have the concept of permalinks
#
[Raphael_Luckom] Maybe I can ask my question a different way. If I give you an arbitrary HTML document, is there anything that will tell you whether it includes a "blog post" as far as RSS or atom is concerned?
#
aaronpk not really, because rss/atom don't really care about html documents
#
[Raphael_Luckom] right, that's what I wanted to make sure I understood
#
aaronpk they exist outside of html
#
[Raphael_Luckom] thanks!
#
aaronpk which is why we call them side files
#
[KevinMarks] Inside the description tags is a the content of the html from the post (or a summary if you're unlucky, because RSS doesn't distinguish summary from content)
#
[KevinMarks] This will be escaped in an annoying way, but it can have microformats in.
[tw2113_Slack_] joined the channel
#
[Raphael_Luckom] yeah also I'm also noticing that hugo is only putting the first paragraph or so into the feed...that seems like something I should fix.
#
aaronpk alright... monocle and aperture are updated now too
#
aaronpk and teacup too. that was a trick because of how old it is
#
aaronpk well that's about all for today! now I wait to hear if anyone has trouble signing in to stuff 😂
[tantek] joined the channel
#
[tantek] [Raphael_Luckom] do you mean is there a way to detect blog post(s) on an HTML page?
geoffo, KartikPrabhu, ethanyoo and jjuran joined the channel
#
aaronpk hmm how much do I care about php 5.6 support
[tw2113_Slack_] joined the channel
#
[tw2113_Slack_] for what product?
#
aaronpk IndieAuth client library
#
aaronpk it's a minor enough thing I should just switch back to the old syntax
#
Ruxton whats ur minimum going tobe if not 5.6? 5.7?
#
Ruxton err 7 even/
#
Ruxton 5.6 hasnt been supported for 2 years, php 8.0 just released.. seems silly to hold onto it. If you keep supporting it, people wont move forward. (the PHP3->4 problem all over again)
#
[tw2113_Slack_] i vote bump to the lowest still-supported version
#
[tw2113_Slack_] which looks to be 7.3 as of 4 days from now
KartikPrabhu, [tantek], nickodd, [KevinMarks], schmudde, themaxdavitt, jeremych_ and [eddie] joined the channel
#
[eddie] My website changed so much over the last couple years both in the types of data I post but also I went from Jekyll to a dynamic site. I’m try to get my website back to life (after server issues) so I’ve been spending some free time this week getting my site’s content updated to a consistent and maintainable format. Instead over the last couple of years as I’ve added features and changed formats I’ve just written a bunch of if statement
#
[eddie] code which makes continuing changes and improvements very difficult. So far things are going well. I’m looking forward to all my posts being in a unified file format shortly 😄 it’ll make life much easier as I get my site re-launched
#
superkuh The most stable, future proof format is .html.
#
[KevinMarks] ideally with microformats in so you can parse it back into structured data if you need to.
#
[KevinMarks] (write tests for this)
schmudde joined the channel
#
petermolnar superkuh: or plain text.
#
petermolnar I'm still fascinated by things like http://www.textfiles.com/magazines/LOD/lod-1 ; plain text, from 1987, perfectly readable, nicely organised.
jeremycherfas and [Raphael_Luckom] joined the channel
#
[Raphael_Luckom] [tantek] no, I think I was trying to get out from under a misunderstanding of how feeds work. I remember using a feed reader a few years ago, and I remembered reading whole articles in the reader. So when I saw that my `index.xml` file only included the first paragraphs of the posts on my site, my first assumption was that the _feed reader_ was then following the link in the feed and displaying the content of the linked resource. The
#
[Raphael_Luckom] other people in this channel helped me realize that that wasn't right. But it was weird because I don't think the questions I was asking really made sense to anyone who understood how it worked, so their answers didn't really make sense to me until I got it.
#
[KevinMarks] that distinction is one of the key reasons Atom was different from RSS - RSS was originally just for summaries (called `description`) but people used it for full posts too, whereas Atom defined `summary` and `content` as distinct elements.
#
[Raphael_Luckom] Another piece of the puzzle, thanks!
[jgmac1106] and geoffo joined the channel
#
[eddie] Regarding the HTML format above. Yes, that is true. However there is metadata that CMSs need that you don’t want server to the browser for backend logic.
schmudde joined the channel
#
petermolnar how would/does a non-blog (read: non reverse chronological) site (eg. digital garden) push updates?
[snarfed] joined the channel
#
[snarfed] petermolnar: wikis probably have lots of prior art there
#
[jgmac1106] The newsletter does a decent job of letting me know of updates, though still technically chronological
#
[jgmac1106] Petermolnar Didn't sites used to say, "sign up to find out when this site changes" was that all RSS?
#
[jgmac1106] Notify me of change is chronological. Forums let you get notified of new topics and/or any posting
#
[jgmac1106] I use the approach, "I don't care if you find it but cool if you do" have dozens of pages outside my feed
[tantek] joined the channel
#
sebbu Zegnat, you there ?
#
sebbu Error: Faulty Request
#
sebbu There was an error with the request. The "scope" field must be used with code requests.
#
sebbu didn't had an issue a few days to a week ago, still using selfauth / mintoken
#
aaronpk are you logging in to something i just updated yesterday?
#
sebbu indielogin.com
#
sebbu so yes ?
#
aaronpk yeah i just updated that yesterday
#
aaronpk i thought zeug at had already updated selfauth to reflect the latest spec though
#
aaronpk lol autocorrect
#
aaronpk Zegnat not zeug
#
sebbu i remember 2 fix (also in open pull requests) but they're older than that
#
vilhalmer hooray, all the implemented parts of my indieauth server are up to date
[schmarty] joined the channel
#
[schmarty] aaronpk: having trouble logging into monocle with https://martymcgui.re/ (which uses indieauth.com)
#
[schmarty] tried with a new private window, so no cookies etc. i get successfully redirected to indieauth.com, can authenticate there, but on redirect back i end up at monocle.p3k.io/login
#
[schmarty] (fwiw i auth'd with PGP)
#
aaronpk uhoh
#
[schmarty] let me see the redirects to see if i can figure out anything more useful
#
aaronpk i wonder if indieauth.com is doing some really old behavior that i just broke in monocle
[arush] joined the channel
#
[schmarty] redirect back to monocle.p3k.io/login/callback includes code, me, and state
#
aaronpk i forgot to look at indieauth.com when i was going through all the apps yesterday, let me take a look
#
[schmarty] can't see the request in the backend of monocle redeeming the code of course
#
[schmarty] but monocle 302'd that request in 146ms so whatever is going on it resolved _quickly_ 😂
#
aaronpk some day i'm going to eventually get rid of indieauth.com :sigh:
#
aaronpk ok confirmed my local copy of monocle doesn't work with indieauth.com
#
[schmarty] wheeee so glad i tried that with a private window instead of signing out of my active session 😅
#
aaronpk ooh indieauth.com is still expecting the "me" parameter in the POST
#
aaronpk also something is wrong with the client library because it should have bubbled up that error to the UI in monocle
#
Zegnat sebbu, aaronpk: https://github.com/Inklings-io/selfauth/pull/46 review required to get selfauth in order ...
#
Loqi [Zegnat] #46 Remove checks based on response_type
#
aaronpk looks good to me but i haven't actually run that
#
aaronpk alright monocle is updated to show that error message now
#
aaronpk guess i should update indieauth.com
#
aaronpk the question is...do i even have a current ruby dev environment
#
aaronpk oh! crap it's the token endpoint
schmudde joined the channel
#
aaronpk lol Zegnat remember how we were like "tokens.indieauth.com is the only thing that'll break if we remove the 'me' from the access token request"
#
Zegnat Hahahaha
#
Zegnat Yes
#
Loqi rofl
#
Zegnat Pretty sure we documented that :P
#
Zegnat I guess we documented it, went ahead with all the spec things because it was not important enough, and never actually stopped to think about what could replace it? Hehe
#
aaronpk https://github.com/indieweb/indieauth/issues/19#issuecomment-399776437
#
aaronpk "The workaround for shared token endpoints is to have a per-user token endpoint URL, like tokens.indieauth.com/user.example.com/token..."
#
Loqi [aaronpk] In order to continue the OAuth compatibility, this would also mean that the code exchange (token request) step would need to work without the `me` parameter as well. This breaks the ability to use a shared token endpoint between users, since the ...
#
aaronpk from 2018 lol
#
aaronpk there's no database and i'm not even logging anything in tokens.indieauth.com so i can't even tell how many different authorization endpoints it's actually being used with
#
[schmarty] hahaha hmm. i am not sure what to do about the above? sounds like it's not possible to fix tokens.indieweb.org quickly. 😬
#
aaronpk i have a plan
#
[schmarty] switch to a new personal token endpoint?
#
[schmarty] 😂 amazing
#
aaronpk haha
#
aaronpk i'm going to first add logging to see if anyone is using it with any authorization endpoint other than indieauth.com
#
aaronpk i'm assuming the answer will be no
#
aaronpk i'm going to also make it default to just use indieauth.com to make it more tightly coupled. if that breaks anyone's stuff then i'll do the per-user token endpoint thing i mentioned in the issue
#
[schmarty] ah, and in the case where it's indieauth.com you can do some kind of private comms between them
#
aaronpk nothing really changes in that case, it just skips the discovery step
#
aaronpk :sigh: tokens.indieauth.com is using such an old version of slim that it won't run in php 7.4
#
sebbu Your IndieAuth server did not return a valid response.
#
sebbu Error Details
#
sebbu Response Code: 307
#
sebbu Maximum (8) redirects followed
#
sebbu hum
#
sebbu weird
#
sebbu indieauth.com and indieauth-client still works
#
[schmarty] i'm grimacing at frameworks these days
#
sebbu i'll try updating indieauth-client
#
aaronpk yeah i feel like laravel does a good job of providing so many features that you actually want to use it, but a lot of these frameworks feel like just a bunch of fluff to make things complicated for no reason
#
Zegnat This is my framework, 8-ish dependencies that together give me full extendibility of how I handle requests, as well as offer routing, testing, and linting: https://github.com/Zegnat/php-website-starter
#
Loqi [Zegnat] php-website-starter: My minimum viable setup for starting a PHP project.
#
Zegnat Big frameworks are just a pain to keep updated. Especially when you think something might be long running (like tokens.indieauth.com)
#
aaronpk that's why i chose this one, called "slim" 😂
#
Zegnat I used Slim for a bit. But found it was still fairly opinionated about what it shipped and did not ship. Although that was before they switched to kinda being like my repo in that they conformed to the PHP FIG HTTP objects
#
aaronpk [schmarty]: i just pushed a fix! try again!
#
[schmarty] !
#
aaronpk it should default to indieauth.com if the client doesn't provide the "me" now, which seems like a safe way to handle this migration
#
[schmarty] SUCCESS
#
aaronpk nice
#
[schmarty] aaronpk++ thanks so much for the quick fix
#
Loqi aaronpk has 71 karma in this channel over the last year (231 in all channels)
#
aaronpk i'm also adding logging so i can see which authorization endpoints people actually use with this
#
[schmarty] those will be interesting stats to know!
[Raphael_Luckom], leg and nickodd joined the channel
#
sebbu welp i broke my auth trying to fix it, well i'll fix it tommorow
#
[Raphael_Luckom] My favorite James Mickens quote: "I have no tools because I've broken my tools with my tools." 😄
geoffo joined the channel
#
sebbu mutual self-destruction
#
sebbu easiest fix, disabling the endpoints :)
[KevinMarks] and strugee joined the channel; nickodd left the channel
#
lahacker what is sqlite
#
Loqi SQLite is a relational database that stores an entire database in a single file https://indieweb.org/SQLite
#
lahacker sqlite << https://www.sqlite.org/json1.html
#
Loqi ok, I added "https://www.sqlite.org/json1.html" to the "See Also" section of /SQLite https://indieweb.org/wiki/index.php?diff=73715&oldid=52613
#
lahacker sqlite << https://dgl.cx/2020/06/sqlite-json-support
#
Loqi ok, I added "https://dgl.cx/2020/06/sqlite-json-support" to the "See Also" section of /SQLite https://indieweb.org/wiki/index.php?diff=73716&oldid=73715
#
lahacker if you were ever hesitant to store JSON blobs in the database for lack of indexing (sorting, joining, ..) see ^
[eddie] joined the channel
#
[eddie] Oh nice!
[jeremycherfas], geoffo, [tantek] and [manton] joined the channel
#
jacky this is dope
#
jacky I could effectively just use my database for everything with this lol
#
jacky (one file to manage)
rhiaro joined the channel
#
lahacker i'll have to give it a proper test but i think this'll help me off the fence wrt primary storage medium; still going to keep a git repo of flat files
#
[tantek] performance << advocacy and a user-submitted light-weight website leaderboard of sorts: https://1mb.club/
#
Loqi ok, I added "advocacy and a user-submitted light-weight website leaderboard of sorts: https://1mb.club/" to the "See Also" section of /performance https://indieweb.org/wiki/index.php?diff=73720&oldid=71264
#
[tantek] jacky noted https://nojs.club/ and it made me wonder, is there a way (like with ACCEPT headers maybe?) to explicitly request a web server to NOT send you any JS at all (including no inline script tags in HTML) ?
#
[tantek] petermolnar, speaking of sites providing "updates" that don't have a reverse chronological display format, note that https://1mb.club/ has an RSS feed link at the bottom.
[Raphael_Luckom] joined the channel
#
[Raphael_Luckom] Would the rationale for that no-js header just be that it would keep the request size down? Otherwise, wouldn't it be easier to do on the client like it is now?
#
[tantek] It would be a collaborative approach rather than the antagonistic approach(es) that require doing all the work "on the client" as it were
#
[Raphael_Luckom] that wouldn't be doable on a static site though, unless you made an apache / nginx / etc webserver aware of HTML semantics in a way that I don't think exists yet.
#
[Raphael_Luckom] I guess you could make everything geerate js and no-js versions
#
[tantek] that's that meta http-equiv is for!
#
[tantek] 🙂
#
[Raphael_Luckom] what is meta http-equiv
#
Loqi It looks like we don't have a page for "meta http-equiv" yet. Would you like to create it? (Or just say "meta http-equiv is ____", a sentence describing the term)
alex11, [schmarty] and [schmarty]1 joined the channel
#
[tantek] [Raphael_Luckom] see https://en.wikipedia.org/wiki/Meta_element#HTTP_message_headers
shoesNsocks and [KevinMarks] joined the channel
#
[KevinMarks] what is baked data?
#
Loqi It looks like we don't have a page for "baked data" yet. Would you like to create it? (Or just say "baked data is ____", a sentence describing the term)
#
[KevinMarks] did we like that term for Simon's sqlite publishing model?
#
[tantek] just added https://indieweb.org/mute#Alternative_to_multiple_feeds for folks if they want to continue on the discussion earlier this week
KartikPrabhu joined the channel
#
[tantek] ^ attempted to describe the high-level motivation for pursuing "mute" as an approach instead of multiple feed files, LMK if that makes sense to folks
#
[KevinMarks] like Kyle making quiettime to ignore me while I'm livetweeting things
#
[tantek] that's a good example
#
[KevinMarks] though when I'm ding that there is a hashtag to mute too
#
[tantek] a different hashtag for each livetweeting session but yeah
#
[tantek] rather than having to explicitly mute every time you're livetweeting, seems like the user would want some sort of frequency / pattern detector for auto-muting all livetweeting (maybe after letting an initial one through)
#
[KevinMarks] I may have been early warning of a hashtag storm too
#
[tantek] tagstorming?
#
jamietanna[m] Given the chats the last few days, do folks have things they do on their IndieAuth consent screens that they think are particularly cool / useful that others may want to try?
#
jacky ^ interested in the above as well
#
jacky I think the most notable thing of mine is that it resolves h-app info and stick that info to any content created with it (at least for micropub, I'd love to track what I've done to a microsub via a specific app)
#
jacky ^ that last bit is part of a larger goal of mine around adjusting content based on the tool I use
strugee joined the channel
#
[tantek] what is using
#
Loqi using is the name of an authoring and/or publishing application, or can also be a page of tools and services that someone uses or likes to use https://indieweb.org/using
#
[tantek] ^ jacky, the intent was to capture that "via a specific app" there, as I think it was Twitter that originally showed on tweets a "using " line with the name of the app.
#
[tantek] ah found it! https://indieweb.org/giving-credit#Crediting_Applications
#
Zegnat jamietanna[m]: not yet, but maybe after this weekend, the ability to set my profile information on the fly for the new profile scope
#
[tantek] thanks to your prompting jacky, added myself to the IndieWeb examples there
#
sebbu lahacker, i also remember utf8 and table constraints (like checks or foreign keys) not being activated by default with sqlite
#
[tantek] yo is there a description for the practice of using dated permalinks that imply they're in the future? e.g. 2020-11-30 hasn't happened yet, and yet: https://www.newyorker.com/magazine/2020/11/30/how-venture-capitalists-are-deforming-capitalism
#
[tantek] should this (using date-based permalink URLs in the future) be a practice we explicitly document / recommend against? perhaps in /URL_design ? opinions?