#dev 2023-07-31
2023-07-31 UTC
tei_1 and IWSlackGateway joined the channel
# 
benatkin I kinda like this one, but it seems to lack implementations in other programming languages. https://github.com/ron-rs/ron
# benatkin After the latest Chrome privacy invasion, "Web Environment Integrity", I started running off to Firefox again. I wish Firefox scored high on this: https://www.bryanbraun.com/2021/03/24/infinitely-nested-iframes/
IWSlackGateway, gRegor, vladimyr__, tei_, tei_1 and [tantek] joined the channel
tei_1, eitilt, petermolnar and [aciccarello] joined the channel
# 
[aciccarello] [snarfed] I'm getting a bad cert error when trying to log into bridgy fed. I moved domain registrars this week but the certificate should be updated now. Is something wrong with my site?
# [aciccarello] I had some issue with my non-www domain which could be related. I thought that was fixed though. BF Error says "Couldn't connect to https://www.ciccarello.me/: 502 Bad Gateway: HTTPSConnectionPool(host='http://ciccarello.me', port=443):"
gRegor, [capjamesg], [KevinMarks] and tei_1 joined the channel
# 
[KevinMarks] You may need to update any map tile links that use Stamen - also they have vector tiles now https://stamen.com/here-comes-the-future-of-stamen-maps/
tei_ and [schmarty] joined the channel
# 
[schmarty] Oh wow. Thanks for the heads-up, KevinMarks++
[Jo], shreyan[m], [0x3b0b], vladimyr, movedtoshreyanbe, bkil, jengaio[m], olaf[m], JaeItIts[m], kandr3s and benatkin joined the channel
# capjamesg [tantek] While I was working on my links project yesterday, I saw the webmention for the post you wrote on venues.
# capjamesg What about a PHP page that takes in an OSM resource, extracts the key data (opening hours, location, etc.), and marks it up with mf2?
# capjamesg I am not content with the way I do checkins right now. Breakfast and Coffee is much closer to the UX I want, but it is for community checkins and aggregating reviews.
[Caleb_Hearth] joined the channel
# [Caleb_Hearth] #lazyweb a bit but are there parsing rules for pulling the u-photo out of a <picture> tag and/or best practices around tagging u-photo? i.e., does it make sense to tag all <source> tags, just the <img> tag, a specific <src> that might be better than the <img src>, etc.?
# capjamesg [KevinMarks] I want the Inbound Links panel to fall back to the Common Crawl link graph.
# capjamesg The link txt file is 24 GB. Then I'd have to mung them into a JSON dict for fast querying. I assume there is no way I can load that without 24 GB+ RAM?
# capjamesg What can I do here? Shard the data by first / second / third letters then do a read for each request?
# capjamesg Ooh! Or a database!
# [KevinMarks] That feels like the kind of size where a document db might make sense, like AWS DynamoDB or Google Firestore. Or you could build it as a db cache over the text file if you don't mind the first hit being a grep of the text file.
# [KevinMarks] How many separate urls does it index? A regular sql db might work - then index size is what wants to be in ram.
# [Caleb_Hearth] ↩️ I think I’ll pop it on <img> based on existing HTML semantics: alt, height, width, title, etc. will use the <img> for any <source> the browser selects.
# [Caleb_Hearth] ↩️ Just means it’ll be picking a potentially too-large image I guess.
# capjamesg Okay so I could handle 25GB on DynamoDB's free tier.
# capjamesg Could I just put these in a Postgres DB though?
[Jo] joined the channel
# [KevinMarks] You could, though I'm not sure about ram limits needed for that size
# [KevinMarks] A lookup by a single key seems like a good fit for the DynamoDB kind of document db. You could potentially stash several kinds of parsed preview in the 400kb record limit.
[snarfed] joined the channel
# capjamesg Firestore looks good too.
eitilt and btrem joined the channel
# [aciccarello] [snarfed] I get the error when trying to log in to https://fed.brid.gy/web-site with https://www.ciccarello.me/
# [aciccarello] This happens before any kind of indieauth flow client side
tei_ joined the channel
jonnybarnes joined the channel
# [aciccarello] Thanks as always for the stellar support! snarfed++
# [snarfed] [aciccarello] welcome! looks like this is due to BF special casing and dropping the www subdomain. it does that first, so it's trying to fetch https://ciccarello.me/ , which serves a mismatched SSL cert for *.http://netlify.app
# [aciccarello] Okay, I'll try to see what's going on with that cert.
tei_, gRegor, vilhalmer and kushal joined the channel
# [aciccarello] I found where to update the certificate on netlify's site and it's working now. Thanks!
paulrobertlloyd joined the channel
# paulrobertlloyd 👋 Am fixing Indiekit’s Micropub endpoint to better deal with update requests that result in no changes being made. This can happen for a number of reasons, such as when using a post editing interface and submitting the form without making any changes. What should the endpoint response be? Am thinking the same as a successful update request (200 OK), but with a message that says nothing was updated, for example: ‘Post at
# paulrobertlloyd https://website.example/my-great-post remains unchanged as no updates were made‘. Here’s the change I’m considering: https://github.com/getindiekit/indiekit/pull/641
# [snarfed] real creates/updates should return HTTP 201, right? https://micropub.spec.indieweb.org/#response
[tw2113_Slack_], [campegg], tei_1 and paulrobertlloyd joined the channel
# paulrobertlloyd Yeah, I was looking at 204 as an option. Maybe its a good choice as its distinct from the the other responses while still be reasonably accurate.
geoffo joined the channel
# [aciccarello] Hi paulrobertlloyd! I'm still looking at getting Indiekit up on my site. I've been playing around with it locally.
# [aciccarello] I'm almost done with the metaformats work on the node mf2 parser too.
paulrobertlloyd joined the channel
# paulrobertlloyd (Ooooh, but if you use 204, you can’t return a message (as no content!). Looks like 200 is the way to go)
# paulrobertlloyd (Or maybe I deal with 204 on consuming side of things… 🤔)
# paulrobertlloyd (Or maybe I deal with 204 on consuming side of things… more fiddling required! 🤔)
# paulrobertlloyd @aciccarello oh great news - was taking a look at the mf2 stuff last night, but decided I’ve got to stop distracting myself from fixing Indiekit’s syndication issues!
# 
Loqi It looks like we don't have a page for "204" yet. Would you like to create it? (Or just say "204 is ____", a sentence describing the term)
tei_1 joined the channel
# [aciccarello] I reached out to the mf2utiljs library maintainer to see if they would be open to community involvement but haven't heard a response https://github.com/drivet/mf2utiljs/issues/1
# [aciccarello] Last night I was noticing some issues with the `wm` node package so I'm also trying to figure out how to move forward with that.
# [aciccarello] Too many things to think about 😅
paulrobertlloyd joined the channel
# paulrobertlloyd tantek good idea. interestingly, looking at https://indieweb.org/Micropub#Handling_a_micropub_request, I see that it says "If the update was successful and didn’t cause the content’s URL to change, return an empty HTTP 204 response". This contradicts the spec which says "the server MUST respond with 200 or 204, depending on whether the response body has content. No body is required in the response, but the response MAY conta
# paulrobertlloyd JSON object describing the changes that were made"
paulrobertlloyd joined the channel
# paulrobertlloyd what is 204
# Loqi 204 is an HTTP status code used to indicate that a request has succeeded, but that the client doesn’t need to navigate away from its current page https://indieweb.org/204
# paulrobertlloyd ☝️ added a page for HTTP 204, and added a section highlighting how the Micropub spec recommends using success status codes
# paulrobertlloyd Think I’m going to return a 200 with note saying nothing has changed, and file a bug that removes the assumption that a Micropub action response will always include JSON
# paulrobertlloyd (a bug against Indiekit)
geoffo and tei_ joined the channel
# 
capjamesg [KevinMarks] bsky.link was linked on https://arstechnica.com/tech-policy/2023/07/musk-annoys-twitter-users-by-capping-number-of-tweets-they-can-view-each-day/
IWSlackGateway and tei_ joined the channel
[tantek] joined the channel
[snarfed] and tei_ joined the channel