stevestreza, wagle, ichoquo0Aigh9ie and [eddie] joined the channel
#[eddie]!tell GWG I change it a lot, and don’t have much data on the posts, so I find it easier to just have the homepage be more visual and have my feed be a separate page that is linked to
#[eddie]!tell GWG once I feel like I nail a homepage design I really like I might spend the time to think about marking it up, but for now I find linking elsewhere the easiest
#LoqiGWG: [eddie] left you a message 1 minute ago: I change it a lot, and don’t have much data on the posts, so I find it easier to just have the homepage be more visual and have my feed be a separate page that is linked to
#@frankmeeuwsen↩️ Je kunt eens kijken of het mogelijk is om webmentions te ondersteunen. Een W3C standaard waarmee je onder andere Twitter commentaar kunt importeren op je site en zo de discussie duurzamer op je eigen domein houdt. (twitter.com/_/status/1153896002006528002)
#sknebelRe the fragment parsing discussion from earlier: I think x-ray just cuts the HTML out of the page, so you loose authorship that relies on other parts of the page etc
[Rose] joined the channel
#[Rose]So, today I found out certbot didn't autorenew some domains... And I can't login to monocle because my aperture isn't https 😛
#ZegnatIs that an arbitrary limitation of Monocle?
#Zegnatsknebel: yes. Actually, I think that may be a custom method in php-mf2 that does that.
#[Rose]I think it was because of my nginx setup - it was telling it to use SSL and the cert it was delivering was expired. A server reboot fixed it, so something was probably stuck in nginx
#[Rose](And to be clear, I approve of Monocle saying "uhh, nope!" in response to an expired SSL cert)
#ZegnatNot sure if that was a specific decision on Monocle’s part. I think default curl will bail out unless you tell it specifically not to verify the cert
#jgmac1106So currently when ever I add a new article to my h-feed, I manuallyhead over to granary, paste in the the link, and then cpy and paste the xml for rss and paste that back on my website. I know some people have automated this. Would doing so require a lot of prior knowledge and time? Anyone make any tutorials?
#Zegnatjgmac1106: I think some people just link to the granary permalink for their homepage
#jgmac1106I did not know I could do that, I just go read some documentation…and slowly back away
#ZegnatI wonder if you can make the case for putting a redirect on your own site, so people can subscribe to jgregorymcverry.com/rss and it will redirect to granary. Then, theoretically, you can switch away from granary without having to tell everyone that they need to resubscribe to you
#jgmac1106In module two folks will add an h-feed of posts, I want to include directions on the easiest way to also add an xml file
Guest2949 joined the channel
#jgmac1106I thought this was gonna be a 1-2 day project but each lesson taking me about half day to write, I figure IndieWeb 101 would take you through Module 1, IndieWeb 201 would be Module 2 and three and four being optional graduate programs
#ZegnatThat will be one of the discussion points, I guess. Should the mf2 parser ever care about the fragment or should the consumer of the parser output do that
#ZegnatBut if the consumer needs to do it, it needs to keep the original HTML around, because the IDs may not be included in the mf2 parser output?
#sknebelThat's why we added IDs to the parser output
#ZegnatYes, but that will only work if the ID is on the same element as the root h-* declaration. Though maybe we do not want to support other cases?
[tonz], [Rose] and [tantek] joined the channel
#[tantek]we already answered that question back in the mf1 days — the parser should handle the fragment because it can do so more efficiently
#[tantek]instead of the consuming code that calls the parser
#[tantek]and I am also against implied h-feed in the parser (what KartikPrabhu said)
#[tantek]I feel we've been seeing an uptick in these kinds of tweets with folks testing webmentions
[renem] joined the channel
#[renem]Since I host my blog via Netlify and want to implement #bridgy-fed I need to write the redirect rules for the ./well-known/ URLs. Problem ist, that redirect rules on Netlify are possible but can't simply forward query parameters as well. Each redirect also needs the query parameter that will be redirected. Can anyone explain me if there is a list with parameters that are necessary? like the "resource" one? any others? Couldn't find a list/spec.
[schmarty], Kaja, jackjamieson, strugee and [Zegnat] joined the channel; jgmac1106 left the channel
#[Zegnat]I am not sure you should expect any query string on that request at all, [renem]
#[renem][Zegnat] I'm following the bridgy-fed setup and there is a quote
#[renem]> Finally, configure your web site to redirect these URL paths to the same paths on https://fed.brid.gy/, *including query parameters*:
#sknebelprobably the webfinger spec is the reference you want...
#[Zegnat]Yeah, I am not sure if there is any reason for that. host-name for one is specifically a static file, AFAIK.
#[Zegnat]You may even be able to copy the one that exists on fed.brid.gy and serve it yourself from netlify. Though of course then you risk being out of date if fed.brid.gy updates it
#sknebelzegnat afaik it asks e.g. for the specific user name in a query oaram
#[snarfed][renem] currently the only query param bridgy fed needs for `/.well-known/webfinger` is `resource`. if you add that, it should work
[zach] joined the channel
#[renem][Zegnat] I’ll try that out and report back if Netlify will work that way. Is there a PR possible to update the docs regarding Netlify redirects? Would write something but don’t know where to publish like [tantek] suggested.
#Loqi[[snarfed]] (we often abbreviate wiki pages here with just the path, eg /Netlify)
#GWG[tantek]: When I asked about where to document things that affected xray and Parse This, you suggested the mf2 parsing algorithm. If implied h-feed algorithms don't belong there... should there be a home for the mf2 consuming algorithms?
sknebel_ and Zegnat joined the channel
#[tantek]"where to document things that affected xray and Parse This" - depends on the specific thing, in this case, I just went with answering the first concrete example, fragments
#[tantek]GWG, I think you need to start with explaining the specific consuming code use-cases for "implied h-feed algorithms"
#GWGI personally need to be able to understand what a page means, and everyone speaks a different language, so I need to convert into terms I understand.
#[tantek]e.g. some of the indiewebify.me features are developer features
#[eddie]!tell [cleverdevil] I'm setting up a blog for my wife's business and the goal is to use serverless as much as possible for scaling. The first step is setting up a Micropub endpoint to create/edit the blog posts. It's going to be a pretty simple standard blog/microblog. Articles, Note, and Photo posts. In your experiments what was the best database for storing posts at Amazon?
#[zach][eddie] I know that the python based cms he was playing with for lambda he ended up using rds/aurora
#[zach]if I recall correctly he was saying that dynamodb can be difficult for some part
#[eddie]ahh gotcha. That's good to know. Thanks [zach] 🙂
#j4y_funabashi[m]DynamoDB avoids long cold starts because it doesn't live inside the vpc
#j4y_funabashi[m]But it does take a change in mindset if you are used to modelling things relationally
#[zach]yeah, I still want to try and use it for a serverless micropub enabled site just because the costs that come with RDS
#j4y_funabashi[m]I use s3 for session data in okami and it works surprisingly well if you only need key/value queries
jackjami_ joined the channel
#j4y_funabashi[m]One thing to note, if you use api gateway in front of your lambda you can't fully support micropub as it doesn't support square bracket array notation in query strings.
#j4y_funabashi[m]I use containers behind an alb and the file size limits definitely don't exist (not implemented square bracket array notation)
#[snarfed]man all of this ^ is really deep for small personal/business web sites, which rarely have big scaling or flash traffic needs. if the serverless motivation is simplicity or scaling, might be worth taking a step back at some point and reevaluating
#[snarfed](lambda is also a pretty limited, low level serverless product. many others are better, higher level, and easier to use)
#ZegnatI have noticed that too. Some people at work were looking into lambdas (for scaling) but before everyone figured out how state would be handled, or authorization of loged in users, etc, we were basically just sketching out a good old vps solution again
#[snarfed]if you really do want to write a lot of your own code for a small web site, serverless can definitely be great! but others like heroku, netlify, glitch, zeit, etc are _way_ easier and more appropriate than lambda
#[snarfed]and shared hosting is still a good option, esp for PHP
#[snarfed]even VPS is probably overkill in many cases, too much sysadmin tax
#ZegnatI wonder at how many visitors the usual cutoff is for a shared hosting
#[eddie]snarfed: Fair enough, other options might be more simpler than lambda. I think there are two major goals: not wanting to spend money on resources we're not using (not paying monthly for a VPS when there is very little traffic, and two being able to scale.
#[snarfed]sure. people usually overestimate their scaling and traffic growth needs, even flash traffic, so i'd think hard about that req't realistically. and also CDNs etc are even better for scaling.
#[snarfed]also will your wife's site have a ton of custom code? just a little? none? if the latter, other options will be tuned for whatever CMS you use, so they'll still probably be better than generic serverless
#[snarfed]having said all that, if part of the point is trying new things, go for it! just depends on what you want out of the project
#[eddie]So the site will have some static pages, but it will also have a subscription component/account portal that will all be custom code using Stripe
#[snarfed]tons of existing services for the custom code side. if the goal isn't for you personally to play with new code/tools, definitely consider them
#[snarfed]oof yes but again S3 is really low level. other static site hosts are way better products
#[eddie]Yeah, the tricky thing is that places like Shopify, etc, like to charge % on top of the credit card processing, etc. Maybe it's worth it? But as I'm also an engineer since the business is all centered around profit on subscription costs, every percent that can be removed from overhead is pretty big. But I guess I do need to factor in how much money we'd be saving based on how much effort it would take
#[cleverdevil]Busy in meetings, but to answer your question, [eddie], I’d actually suggest looking at how far you could get just by storing MF2 or JF2 JSON in S3, rather than using a database.
#[cleverdevil]If you have to use a database, RDS is the easiest to deal with.
#[cleverdevil]DynamoDB is a PITA and DocumentDB, while great, is too spendy for a simple personal site.
#Zegnat[cleverdevil]: at what scale would DocumentDV not be “too spendy” anymore?
#[eddie]After looking at some of the options in the above conversation, I have to say while I don't feel like most of them fit, I'm actually thinking Netlify seems promising
#j4y_funabashi[m]So I am building a thing and not sure what to call it. It is kinda the reverse of the current micropub -> static site tools
funwhilelost[m]1 joined the channel
#j4y_funabashi[m]It pages through your micropub GET q=source endpoint and saves the posts as MD files with front matter
#j4y_funabashi[m]I guess it is more like webmention.io but for posts
[KevinMarks] joined the channel
#[KevinMarks]Heroku for node is fairly non-spendy if you want relatively simple set-up compared to lambda.
#[KevinMarks]J4y you could even parse posts for mf2 and then make the md files for Hugo. I was thinking about doing that for my site as it's all manual at the moment
#j4y_funabashi[m][KevinMarks]: yes that is what it does, grabs all the mf2 from your micropub endpoint, converts it to jf2ish and saves as Hugo post files
[grantcodes] joined the channel
#[grantcodes]I planned on making postr a headless micropub endpoint, but haven't got around to writing a module for it yet!
#[KevinMarks]Right, but it could work for a static site too potentially
#j4y_funabashi[m][KevinMarks]: ah yes, the source of the mf2 doesn't have to be a micropub endpoint, it could crawl any public site
#[KevinMarks][renem] if you're the only user you can serve the webfinger page statically on your own site - I think aaronpk does that.
[tantek] joined the channel
#[tantek][aaronpk] Tracking down some interesting IndieAuth interactions and looks like dbryant and I have discovered an inconsistency between how the wiki (IndieLogin) works and how Aperture and OwnYourGram work with signing-in with IndieAuth
#[tantek]In particular, if he tries to use just his domain to sign-in to the wiki (IndieLogin), it (presumably the WP IndieAuth plugin) does ask to authentica him as his domain.
#[tantek]However when he tries to just his domain to sign-in to Aperture *or* OwnYourGram, the IndieAuth authetication screen shows his domain/blog as the identity that is being authenticated
#[tantek]so *something* is working differently with IndieAuth sign-in on those two
#[tantek]the latter two seem to be errantly changing the identity that he's trying to sign in with from his domain to his domain/blog
chrisaldrich and [frank] joined the channel
#[frank][grantcodes] I wanted to take the new Together for a spin but it can't seem to load my Aperture-based subscriptions? I revoked old IndieAuth tokens, but no luck. My subs are at https://aperture.p3k.io/microsub/146. When I login I get the message "error loading channels" and nothing else. Any idea how I can move forward?
#[grantcodes]Hmm have you tried logging out and back in again? That should refresh the tokens now (just fixed today)
#[frank]Weird. I don't have the Yarns plugin active at my WordPress backend but it tried to access that one. Now I've deleted it and it still gives me GraphQL errors in the dev console
#[frank]You want me to open a bug at Github for it?
#[frank]I'm off to bed so will not test much longer now...
#[grantcodes]Go for it. There should be a link in the console to create a auto filled issue
#[grantcodes]Hmm was there another link? That one shouldn't break everything I don't think