#dev 2022-02-16

2022-02-16 UTC
#
[tantek]
but above all I don't want to comingle content with different source/authority/reviewed state so I might have to disentangle it later
angelo_, angelo and barnaby joined the channel
oodani joined the channel
#
[tantek]
whoa that seems like a lot!
#
barnaby
yeah I’m kinda amazed tbh. where are all these applications parsing microformats?!
#
[tantek]
php-mf2++
#
Loqi
php-mf2 has 1 karma over the last year
#
[tantek]
possibly WordPress installs?
#
barnaby
the chart is also hilarious. it got super popular one time back in 2017
#
[tantek]
GWG, do any of the WordPress plugins depend on php-mf2 using packagist and auto-install it?
#
aaronpk
that would also be counting development installs as well as CI builds right?
#
barnaby
yeah, it’s every time it gets installed, so it’s a pretty inflated number
#
[tantek]
that 2017 bump looks like it happened right after IWS 2017
#
[tantek]
what is 2017
#
Loqi
IndieWeb Summit 2017 was 2017-06-24 through 2017-06-25 in Portland, Oregon; the seventh annual gathering for independent web creators of all kinds, from graphic artists, to designers, UX engineers, coders, hackers, to share ideas, actively work on creating for their own personal websites, and build upon each others creations https://indieweb.org/2017
#
barnaby
so every push I make to, e.g. taproot/micropub counts for three installs (one for each version of PHP in the CI)
#
aaronpk
it's still an indicator of activity though!
#
barnaby
absolutely, and it’s always fun to see Internet Numbers go up, even when you know they’re not particularly meaningful xD
#
GWG
Not specifically...
#
[tantek]
barnaby, absolutely, I keep hearing about this "number go up" thing being a good thing 😉
#
@4cm4k1
↩️ i have thought about this, too, as my personal site simmers… have considered webmentions but idk https://indieweb.org/Webmention
(twitter.com/_/status/1493768810913959944)
#
@HongPong
↩️ webmentions are a refined pingback if you will
(twitter.com/_/status/1493772057376661504)
alex11, robo, [Zeina]1, [aciccarello]1, cygnoir[d], angelo, ben_thatmustbeme, [tonz]1, [schmarty]1, sp1ff and [Zeina] joined the channel
#
capjamesg[d]
[James_Van_Dyne] I just pushed v0.2.0 to PyPi. It has been a pleasure working with you on this release!
#
capjamesg[d]
[James_Van_Dyne]++
#
Loqi
[James_Van_Dyne] has 3 karma over the last year
#
capjamesg[d]
IndieWeb Utils is now in v0.2.0! 🎉
#
@robertvanbregt
After some long trial and error, I’ve finally managed to get the Syndication Links plugin working again with http://brid.gy after a migration to another web host. Shoutout to @dshanske (David) for helping out.
(twitter.com/_/status/1493854817307013129)
robo, [Zeina], lagash and [James_Van_Dyne] joined the channel
#
[James_Van_Dyne]
Oh Far out!
#
[James_Van_Dyne]
capjamesg++
#
Loqi
capjamesg has 30 karma in this channel over the last year (76 in all channels)
#
[James_Van_Dyne]
capjamesg: Thanks for making the release and not getting annoyed with all of my change requests 🙂 . I’m looking forward to integrating it with Tanzawa, which I’m sure will result in some more collaboration on the library for fixes / improvements.
#
capjamesg[d]
Ah no worries. I already have a few things I'd like to do next haha.
#
capjamesg[d]
A big one is allowing one to access HTTP headers from a send webmention request, or at least the Location one.
#
capjamesg[d]
But that's for another release 😂
#
capjamesg[d]
[schmarty] Why the term "caturday" doesn't work in IndieWeb Search is so ridiculous.
#
capjamesg[d]
It's almost as if Elasticsearch doesn't like that word 😂
#
sknebel
is it to similar to "saturday" and eliminated as low-signal? :D
#
capjamesg[d]
I'll have to check how your site is indexed [schmarty].
#
capjamesg[d]
Do you ever use the term outside of "#caturday" [schmarty]
tetov-irc and jacky joined the channel
#
jacky
looking at image proxies and found this one https://github.com/h2non/imaginary (still looking for a video proxy)
#
jacky
has lots of bells and whistles
#
Loqi
[h2non] imaginary: Fast, simple, scalable, Docker-ready HTTP microservice for high-level image processing
#
sknebel
hm, not sure video stuff is likely to be a proxy setup
Guest6 joined the channel
hans63us[d] joined the channel
#
sknebel
what do you want it to do? wonder if mediagoblin/peertube/... have useful bits to borrow
#
jacky
tbh I think I was looking for a solution to a problem I didn't have
#
jacky
like the only thing I really needed was prevention of mixed content warnings but I can do that myself using a little web proxy
#
sknebel
ah, in reader?
#
jacky
(and I kind of consider my site to be a reader in a way but read-only)
oodani_, angelo and KartikPrabhu joined the channel
#
[snarfed]
congrats on the new indieweb-util release capjamesg!
#
[snarfed]
I noticed a few things you may want to add to your webmention sending code, just fyi
#
[snarfed]
in discovery, you'll want to check a tags as well as link tags, check for text/* content-type before looking through the response body, and drop fragments from discovered endpoints
#
[snarfed]
also testing on localhost is common (and useful!), so you may want an option to allow that
#
[snarfed]
in send, you probably want to disable following redirects in `requests.post`, since it will downgrade those HTTP requests from POST to GET
#
[snarfed]
(I compared your code with mine briefly, https://github.com/snarfed/webutil/blob/main/webmention.py . yours has lots of useful features!)
#
capjamesg[d]
snarfed PRs are always welcome 🙂 Otherwise, I'm happy to create an Issue for this and come back to it! Thank you for the feedback, there are some things in your message I didn't think about.
Seb[d], jacky and barnaby joined the channel
#
[snarfed]
capjamesg++
#
Loqi
capjamesg has 31 karma in this channel over the last year (77 in all channels)
#
[snarfed]
oh also, the `me` kwarg to send_mention, did you want to check it against source instead of target? https://github.com/capjamesg/indieweb-utils/blob/ee701bb75b066720e78736ea5f91f55a27d4de97/src/indieweb_utils/webmentions/send.py#L44
#
capjamesg[d]
I actually did want to check it against target. The idea is that you can block webmentions that aren't sent to you.
#
capjamesg[d]
That function should probably be expanded to allow for you to set multiple domains.
#
capjamesg[d]
I use that feature to block my receiver from processing any webmentions where target != https://jamesg.blog/*
#
[snarfed]
huh, but it's send, not receive. guess I'm confused about the usage
#
capjamesg[d]
Oh no! Sorry, I got confused there.
#
capjamesg[d]
That definitely isn't right.
#
barnaby
is https://github.com/glennjones/microformat-node still the go-to mf2 parser for node? it’s the one listed on microformats.io, but the codebase hasn’t been updated in years and is likely missing some of the newer features e.g. img alt parsing
#
Loqi
[glennjones] microformat-node: Microformats parser for node.js
#
Loqi
[microformats] microformats-parser: A JavaScript microformats parser for the browser and node.js
#
sknebel
(havent had time to look at it in detail, but its actively supported. putting up a test site for it would be extremely useful)
#
barnaby
thanks! looks like it’s more active than Glenn’s library. might be worth swapping the link and instance on microformats.io (whoever it is who runs that site now?)
#
sknebel
we do ;)
#
Loqi
[microformats] microformats.io: Simple explainer site about Microformats.
#
sknebel
afaik all around it got transferred and at least aaron has access to the heroku behind it
#
aaronpk
yep and it should be set up to auto deploy from github
#
aaronpk
oh wait even better, that one isn't on heroku anymore since it was just HTML
P1000[d] and [Joe_Crawford] joined the channel
#
[tantek]
Yes we should swap to the new JS mf2 parser
#
[tantek]
No one has picked up maintenance on Glenn’s microformatsnode parser
#
[tantek]
For years
#
[tantek]
It's ok to put that in a “Prior work” section (as I think we have for other old/out of date parsers)
#
barnaby
I just started writing a guide to consuming microformats data (the much belated counterpart to https://waterpigs.co.uk/articles/getting-started-with-microformats2/), and want to cover as many “gotchas” as possible — any suggestions for best practises, things to watch out for, useful heuristics or techniques?
#
barnaby
I’ll definitely be poking through aaronpk’s xray codebase for inspiration
#
[tantek]
barnaby, not sure how much a generic “how to consume mf2” makes sense? Makes more sense to have actual human-centric use-case specific docs
#
[tantek]
Especially if you're going to go into best practices and heuristics
#
barnaby
links to good resources for handling untrusted data (e.g. avoiding XSS, unicode issues etc) would also be appreciated
#
barnaby
[tantek]: it’s more about how to deal with canonical mf data structures, although I’m definitely going to mention various use-case specific algorithms such as authorship and representative h-card
#
barnaby
stuff like “never rely on a property value being a plaintext string; if you need a plaintext string and get a struct, use its `value` key”
#
barnaby
which was recently shown to be relevant and little-known by the discussion log outage following php-mf2 0.5 including the new img/alt struct
#
[tantek]
Key stuff to mention that applies in multiple use-cases:
#
[tantek]
• Handling img+alt properly
#
[tantek]
• Handling getting the “value” of a property e.g. when it may or may not have an embedded object as a value (which is technically possible for all properties!)
#
[tantek]
• Handling the first value vs multiple values of a property
#
barnaby
so there are definitely some general principals which are worth documenting separately to use-case-specific algorithms
#
[tantek]
But if you find yourself talking about photo vs logo stop because that's not about mf2 in general
#
barnaby
and I also want to make the whole process more accessible for people just starting out
#
[tantek]
That makes sense
jacky joined the channel
#
barnaby
it’s my blog post and I’ll write what I want in it :P I might mention photo vs logo in a section about dealing with h-cards, but I’m concentrating on general approaches
#
barnaby
first value vs multiple values is especially interesting in the case of u-url parsing, where I’d generally assume that the first URL is the user’s canonical URL, and any others would be secondary e.g. silo profiles, although to really confirm that you’d need other heuristics
#
barnaby
which has some overlap with representative h-card, but isn’t exactly what it does
#
barnaby
same with photo — at least for h-card, most consumers that I know of will only use the first one, but I can imagine use cases for multiple photos. and photo on a h-entry can obviously have multiple, but at least in many silo photo post UIs, the first photo is given precedence (shown larger, first in slideshow etc)
#
[tantek]
Oh I thought you were adding this as content to the wiki. Sure, blog what you want :)
#
[tantek]
Also note that more atomic single topic focused blog posts (that build on each other) tend to be more effective.
#
barnaby
[tantek]: I considered adding it to the wiki but I prefer blogging more informal “how to” content like this, and wanted to follow-up my post for publishing mf2
#
[tantek]
Yup, makes sense to start with a blog post like that to capture a snapshot in time
#
[tantek]
Can selectively add to the wiki from that later
#
barnaby
exactly
#
[tantek]
Since you said “how to” I thought you were going to edit the actual “How To” sections of various pages
#
[tantek]
No worries and we can do that later
#
capjamesg[d]
snarfed Feel free to submit a PR to handle the me issue. Otherwise, I'll get to it soon. It's definitely a priority (but easy!) fix.
jacky, chenghiz_, willnorris and tetov-irc joined the channel