#AngeloGladdingsnarfed tantek using HEAD and an async job queue good enough for now -- can understand why brid.gy breaks spec -- will be sure to document my future webmention caching endeavors
[pfefferle], barpthewire and [gerwitz] joined the channel
#[gerwitz]@sknebel sorry for late reply: in my case I have a Micropub endpoint that is write-only, so valid-to-spec requests to update or delete must throw errors. Of course, I don’t want to imply the request is malformed.
#[gerwitz]…so 400 is the obvious response code, but I’m not suer whether “invalid_request” is appropriate
#cweiskeisn't there some "unsupported http method" status code?
#[gerwitz]🤔 I also felt like 5xx was appropriate, but I do “understand” the request method and it seems contrary to the spirit of Micropub to send a parsable error outside 4xx.
#[gerwitz]i.e. it’s a Micropub error, not an HTTP error.
#cweiskethat error is not limited to the request method.
#cweiskeit's correct for all types of does-not-support-functionality
#[gerwitz]I’m inclined to agree and since you’ve implemented clients I feel emboldened to use 501. 🙂
KevinMarks_ and KevinMarks joined the channel
#Zegnat501 sounds good to me too, but you might have to invent your own response code. Maybe file an issue against the spec, [gerwitz]? Including whatever you end up using.
#ZegnatAs you can see, it submits a source and a target URL to the webmention endpoint. This CMS (Hugo) prefills the target field, and the source filled is then filled by the website visitor.
#ZegnatYou can use exactly that same HTML form, you just have to have a way for your CMS to fill the target URL. And you have to point the form action at your receiving endpoint (https://webmention.io/scripter.co/webmention).
#ZegnatAs you can see, it submits a source and a target URL to the webmention endpoint. This CMS (Hugo) prefills the target field, and the source filled is then filled by the website visitor.
#ZegnatYou can use exactly that same HTML form, you just have to have a way for your CMS to fill the target URL. And you have to point the form action at your receiving endpoint (https://webmention.io/scripter.co/webmention).
#kaushalmodi_Hello, correct, I already looked at that
#ZegnatWebmention endpoints actually accept webmentions in “form encoded” format. So any web browser can submit a form to it, no need to have special sending logic.
#kaushalmodi_But that uses a heroku app as I mentioned in the other channel
kaushalmodi_, KevinMarks and [eddie] joined the channel
#[eddie]j4y_funabashi: wow! I’ve been thinking about a Micropub Lambda. How cool!
j4y_funabashi joined the channel
#petermolnarso... if I grab a few million URLs and send each of them as source, in a loop, through the heroku page, who's webmention endpoint would survive it?
#petermolnarunfortunately nothing has been put to practice so far
#ZegnatI guess that could work. It does stop all those nice URL forms people are putting beneith their comments from working.
KevinMarks joined the channel
#petermolnaryou could whitelist the form on your own site
eli_oat and KevinMarks joined the channel
#ZegnatThat does up the difficulty level of getting into webmentions from adding a few lines of HTML to keeping DNS up to date
#petermolnarwe could leave out dns, in theory: add another meta, say, webmention-spf, into the header, on first fetch, cache it for a while; this only protects from one type of the attacks though
#ZegnatAnd still requires you visit the website at least once to retrieve meta. But might work.
#petermolnarat least it would stay within HTTP, so no obscure, arcane, distasteful DNS knowledge required ;)
[kevinmarks] joined the channel
#[kevinmarks]Mention-tech will check the validity itself before sending ding them in, but that will potentially fetch the page.
#[kevinmarks]So if you post to mentiontech.com it will post through to the actual endpoint, and cache it in case there isn't one.
#j4y_funabashihuzzah! my lambda based micropub endpoint passes all the micropub.rocks test I care about
#LoqiIt looks like we don't have a page for "mentiontech" yet. Would you like to create it? (Or just say "mentiontech is ____", a sentence describing the term)
#LoqiIt looks like we don't have a page for "mentiontech.com" yet. Would you like to create it? (Or just say "mentiontech.com is ____", a sentence describing the term)
#LoqiIt looks like we don't have a page for "serverless" yet. Would you like to create it? (Or just say "serverless is ____", a sentence describing the term)
#[eddie]That is the combination I was recently thinking about!
#[eddie]Awesome, it’s great to see someone already work on implementing that and that it’s not just an idea in my head
#j4y_funabashiI have had the same setup in one form or another across the many iterations of my site (json files + DB cache)
#[eddie]Currently my files are all YAML, but that means lots of conversion to and from JSON/YAML and I find I have more issues manually editing the yaml because of the white space requirements. I always mess up my whitespace, where as with JSON it’s much more rare to have issues and with a JSON linter it’s almost never happens. So I’m thinking when I transition from a static site to a dynamic site, I might convert all my YAML to JSON
#[kevinmarks]as long as you have an editor that is verifying structure
#j4y_funabashiI am converting the code behind j4y.co to lambda/dynamo and also switching domains
#[eddie]kevinmarks: Even without structure verification, I at least, am finding I am better trained to catch missing ,s and {s than missing whitespae
#j4y_funabashiThe files I store in s3 are microformats json, then when I insert to dynamo I convert it to 'jf2'
#[eddie]Just yesterday I edited a YAML file and tried to rebuild my site only to find it broke because of a whitespace problem
AngeloGladding joined the channel
#[eddie]ohhhh, that makes a lot of sense j4y_funabashi. I definitely was thinking mf2 json, but since it’s a one way data flow from s3 -> dynamo, that does seem like a convenient conversion.
#[eddie]What language do you write your lambdas in?
#Zegnatj4y_funabashi, YAML gets not-fun very fast in my experience
#j4y_funabashi[eddie]: Yeah I am liking storing jf2 in the DBcache as it makes all the view code alot simpler
#[eddie]Cool, I’ll take a look around the repos. Are your website views going to be serverless or are you just going to have a server running the view side of things?
#aaronparecki.comedited /timeline (+0) "/* Microsub */ found older draft on my laptop, unsure of start date but this was the oldest date i found" (view diff)
#ZegnatI use a u-author on the feed itself for my feed page, a visible one.
#ZegnatBut it is also rel="author". Not sure why I choose to use both.
#kaushalmodSo with rel="author" added to the u-url in my h-card, I need to add just "<a href="/" class="u-author h-card">Kaushal Modi</a>" in my h-entries. Correct?
#sebselZegnat a rel="author" on the feed or on the entries?
#ZegnatOn the A element that is the u-author for the feed, sebsel
#tantekZegnat in the past Google used to advocate for folks to use rel=author. Then they stopped advocating. And there is some anecdotal evidence that they may (sometimes?) ignore rel=author as well
#ZegnatI doubt any of my markup was written to please Google :P
#ZegnatAlthough Google might be one of the few automated systems actually reading my site
#tantekright, there's no need to - they can read your mf2
#aaronpkkaushalmod: the rel="author" isn't needed at all actually, you just need the matching `u-url` on both your footer h-card as well as inside the h-entry
#sebselaaronpk I wanted to focus on my own frontend for now first, to get to know the thing better.
#sebselEspecially now you've made it so easy :) (they say)
#ZegnatOh, sebsel, there isn’t an “issue with the description of the” edge-case, if you refer to why Aperture isn’t getting my blog. It isn’t strictly an issue with Aperture at all.
#tantekaaronpk /authorship already describes how to for both posts and streams of posts, thus apparently we decided on permalink + streams (home page or separate feeds)
#tantek.comedited /authorship (+423) "update how to publish to not recommend rel=author explicitly, since it was intended for backcompat, and mixed class/rel suggestions are known to be more confusing than just class, note rel-author link detection for backcompat" (view diff)
#kaushalmodAlso I missed "homepage" in "Check your homepage h-card:" on that validation page
#kaushalmodthe h-card validates as being representative on the homepage alright
#ZegnatThe question behind the representative h-card: “Given a URL to a page that represents a person, how do you determine which hCard on that page is the hCard for that person?” - http://microformats.org/wiki/representative-hcard
#LoqiThe representative h-card for a page is an h-card on that page that represents that page, if any, as not all pages are about a person or organization, a page might not have a representative h-card https://indieweb.org/representative_h-card
kaushalmodi joined the channel
#tantekalright I'll give it 24h but I'm already testing the waters with seeing about deprecating meta name=author from HTML, just asked if anyone knows any browser support in the #whatwg channel
#sebselhas this changed? how did this ever work for me before?
#sknebelsebsel: if I'm not mixing up things, only your micropub endpoint is using the verification, so as long as you did the same there or are using a shared database you wouldn't notice
#sebselAh you're right. That must be it. And then I just missed this when I build it.
jjuran joined the channel
#sebselbeh, I don't think I ever implemented it. I should probably go back to rebuilding my site before going on this reader-adventure :(
#tantekin reading http://kimberlyhirsh.com/dear-indieweb/ has anyone else had this problem/challenge? "Posting notes to Facebook and Twitter without weird link previews or my Gravatar popping up"
#Loqi[Kimberly Hirsh] Dear IndieWeb, it may be time to start considering the user, not just the technical spec. by Eli Mellen (eli.li)
I’ve been working on a series of walkthrough posts that outline how to IndieWebify a Wordpress site. I presumed the initial setup would...
#tantekI feel like there's an opportunity here to document a /link-preview#How_to that specifically says how to *not* have a link-preview
j12t, j12t_, KevinMarks and tbbrown joined the channel
#gRegorLoveThat sounds like messed up mf1 in WordPress templates.
#gRegorLoveOr maybe not. Could be OGP finding their gravatar as the only image on a post and showing that in the preview. That would make more sense. I mis-read the comment at first.
#gRegorLoveI have run into this: the image I specified for OGP quite a while ago is smaller than FB's "required" resolution, so link preview finds other larger images on my links and defaults to them instead.
#gRegorLoveAnd that's explicitly using their ogp tag, so that's frustrating.
#gRegorLoveThe UI does give the option to select other images, so I can click an arrow until it shows my icon instead of the larger photo.
#tantekbut that requires manual 'fixing' of POSSEd posts to FB
#gRegorLoveI've been mostly manually POSSE to FB. I haven't tried in a while, but when I was using Bridgy Publish I wasn't always getting backfeed if the post showed a link-preview underneath.
#gRegorLoveSo I paste in my link and remove the link-preview.
#gRegorLoveAnd sometimes, like with that Ars Technica piece, I actually post the entire note text to FB instead of a title+link, so manual is easiest for me currently.
#tantekI'm rethinking all my FB POSSEing in light of all the privacy/CA betrayals
#tantekand frankly even locking down things like wall posting
#kaushalmodiI save that JSON to <site repo>/data/disqus_comments.json and then just loop through it, and output the comments where the page RelPermalink matches that in the db