#dev 2021-10-05

2021-10-05 UTC
KartikPrabhu, hendursaga, akevinhuang, jjuran and gRegor joined the channel
#
capjamesg[d]
I am collecting per-page data on IndieWeb Search now. It might take a while for the data to come in though.
#
capjamesg[d]
Thanks for collating that data [snarfed]!
#
capjamesg[d]
[tantek] My crawl is running pretty actively so I'm happy to pull any other data that you might find interesting (assuming [snarfed] does not already have it, because his dataset is already comprehensive).
#
capjamesg[d]
[snarfed]++
#
Loqi
[snarfed] has 32 karma in this channel over the last year (59 in all channels)
#
[tantek]
very cool capjamesg[d]++
#
Loqi
capjamesg[d] has 11 karma in this channel over the last year (21 in all channels)
#
[tantek]
what's that feed discovery summary index link that you had
#
capjamesg[d]
It's not fully automatic. The list will be updated whenever new records are added to the index. This typically happens once per day, or more often if I myself am doing something.
#
nekr0z
capjamesg[d]: I wonder if your data would allow to create a page that would show "the IndieWeb pulse": something like "the most liked/replied to page this month", "and event with the biggest number of positive RSVP this year" — something along those lines.
#
capjamesg[d]
[tantek] I love your "using BBEdit" tag at the bottom of posts.
#
nekr0z
I understand the crawling is too slow for "today" or "this week", but...
#
capjamesg[d]
I have 2.2 million links in a CSV from the last crawl.
#
capjamesg[d]
All link calculations are scheduled to happen monthly. They take a few hours to run on the 250,000+ URLs currently in the index. But that is very much doable.
hendursa1, ben_thatmustbeme, justache, [fluffy], gerben, [pfefferle] and tetov-irc joined the channel
#
[KevinMarks]
Leaderboard pages are a bit tricky as they are self feeding and a spam magnet, but can be a good opportunity to use Twyman's law to spot the spammy.
[jacky] joined the channel
#
[jacky]
Do we have a way to do like one-off "webmention validation"? I can think of https://webmention.rocks but I think the person looking for something is looking for something that could show what it'd look like on a receiver end
#
[jacky]
I vaguely remember a p3k service doing this indirectly
#
[tantek]
capjamesg[d] oooh good summary!
#
[tantek]
I’m curious what these all mean though. Like “feed”?
#
[tantek]
Are these all rel values?
#
[tantek]
And what’s the denominator? Eg across how many sites?
#
[tantek]
And the text/css one is likely an alternate style sheet, not a feed
#
capjamesg[d]
[tantek] The values are all rel values from <link> tags at the moment.
#
capjamesg[d]
"feed" and "websub" are reserved.
#
capjamesg[d]
My recrawling script needs to know the exact type of feed found so I have a few reserved names.
#
capjamesg[d]
"feed" is a feed found in a <Link> header.
#
capjamesg[d]
HTTP header, that is.
#
[tantek]
Hmm they looked more like “type” values
#
[tantek]
Does http header mean it may also be present in a link tag? Would that be double counting?
#
capjamesg[d]
Sorry, they are type values 🤦
#
capjamesg[d]
Feeds are only counted once.
#
capjamesg[d]
So if jamesg.blog sends a Link HTTP header and has a feed in a <link> tag for jamesg.blog/feed.xml, that will only be put in once.
#
capjamesg[d]
This is still in development 🙂
hs0ucy joined the channel
#
[tantek]
Does “feed” mean there was no type then?
#
capjamesg[d]
Yeah, no explicit type.
#
capjamesg[d]
How do people handle errors on their personal websites?
#
capjamesg[d]
I am running a couple of different services now but there is not an easy way I can track some errors without having to go into a server and read the logs.
#
capjamesg[d]
What is error handling?
#
Loqi
It looks like we don't have a page for "error handling" yet. Would you like to create it? (Or just say "error handling is ____", a sentence describing the term)
kogepan joined the channel
#
nekr0z
<capjamesg[d]> "How do people handle errors on..." <- What exactly do you mean by "handling errors"? I thought Apache did a good enough job of handling my website errors so I didn't have to... ;)
#
capjamesg[d]
I meant tracking errors.
#
capjamesg[d]
Error handling is all good technically. But knowing when errors happen is a bit different.
#
capjamesg[d]
I can see that an app crashed by looking at the logs but I was wondering if anyone had set up some kind of error notification system for their site.
#
aaronpk
re: https://twitter.com/_/status/1445393745202098192 that's a very nice use case that also has a very straightforward solution already. I wonder if it would get any traction if there was a nice UI for typing in a domain and seeing all the rel=me links, and had a little "how to show up here" guide below to encourage more people to put rel=me on their social media links on their home page
#
@zachleat
@theworstdev I want to build an IndieWeb service for this. Enter a domain and it gives you all of their socials BUT WHO HAS THE TIME
(twitter.com/_/status/1445393745202098192)
micahrl[m] and shoesNsocks joined the channel
#
capjamesg[d]
Still in development 🙂
#
capjamesg[d]
Yours is quite long aaronpk 🙂
#
aaronpk
oh nice
#
capjamesg[d]
22 rel=me links picked up haha.
#
aaronpk
oh i think i have one more to add now too 😂
#
petermolnar
capjamesg[d]++
#
Loqi
capjamesg[d] has 12 karma in this channel over the last year (22 in all channels)
#
hs0ucy
Hello,
#
hs0ucy
What do you think of using fragments instead of pages for categories/tags?
#
capjamesg[d]
aaronpk "How to show up here" is now live on the search engine. aaronpk++ for the idea.
#
capjamesg[d]
It's up to you to choose how you represent categories/tags.
#
capjamesg[d]
I actually used that approach a while ago.
#
capjamesg[d]
Because it was much easier to execute on a static site like Jekyll (on which my site is built).
#
hs0ucy
I'm using a custom ssg and I don't want to create a lot of pages for each tag ... I could but, it's a lot of pages :S
#
capjamesg[d]
The disadvantage is that the page starts to become more difficult to navigate as you add new categories.
#
capjamesg[d]
Yeah, that's why I used to have one page like your site does 🙂
#
aaronpk
nice! you should probably link to this instead https://indieweb.org/rel-me
#
capjamesg[d]
My first user-facing change in a little while. Recrawling has taken up too much mental headspace.
#
hs0ucy
capjamesg[d]: Yeah I don't think it's a corner stone for my site, just a nice to have ... people can access to it by clicking a.p-category in post (notes, articles, bookmarks).
#
capjamesg[d]
I think having a category page is better than not having any at all.
#
capjamesg[d]
You can always have "skip to the top" links.
#
capjamesg[d]
So that way people can easily get back to the top of the page.
#
hs0ucy
yep
#
capjamesg[d]
You could have one of the fancy arrows like some websites have or even just a link below each subheading.
#
hs0ucy
capjamesg[d]: yes i'm plannig to do something like that ... thanks for your feedback
#
capjamesg[d]
No worries. It's super easy to do with HTML IDs.
#
capjamesg[d]
I have a link at the top and bottom of every page for accessibility that just links to a HTML ID.
#
capjamesg[d]
"Skip to main content" and "Go back to the top"
#
capjamesg[d]
You could always use <details> and <summary> tags to show and hide links in categories too.
#
hs0ucy
yes I have already visited your site ;)
#
capjamesg[d]
So that way the page is just a list of category names and one has to click on a name to see the categories.
#
capjamesg[d]
I don't know what happens if you link straight to an ID inside a closed <details> tag though.
#
hs0ucy
capjamesg[d]: yes my archives page is like that https://hugo.soucy.cc/archives.html
#
hs0ucy
i would be cool to open a details only with CSS ... with a :target pseudo class for example
#
hs0ucy
.year:target + datails {...open}
#
hs0ucy
I think we can do that only with JS ...
#
capjamesg[d]
I don't even think you need that. HTML opens the details for you 🙂
#
nekr0z
<capjamesg[d]> "Error handling is all good..." <- Oh, mine is a static one. The server emails me if there are issues during a rebuild, other than that there are no errors to expect. Of course, Apache or DNS could fail, but it's too unlikely to bother monitoring ;)
#
capjamesg[d]
nekr0z It's more for IndieWeb Search and a couple of other services I host (webmention endpoint, micropub server / client, etc.)
#
capjamesg[d]
My site is static too 🙂 But I have lots of dynamic sites on subdomains.
#
[snarfed]
capjamesg re tracking errors, the standard is to plug a collector into your code that sends uncaught exceptions somewhere like https://sentry.io/
#
[snarfed]
which aggregates them, alerts you, and has lots of other features
#
hs0ucy
capjamesg[d]: I heard jekill is slow, is it true? how many entries do you have on your web site?
#
capjamesg[d]
Probably around 500
#
capjamesg[d]
It builds in 30 seconds with Netlify.
#
capjamesg[d]
They have some caching system which speeds up deployments.
#
capjamesg[d]
761 right now.
#
capjamesg[d]
[snarfed} Thanks!
#
hs0ucy
capjamesg[d]: I build locally and deploy with rsync ... ~400 entries in ~3sec
#
capjamesg[d]
I might do that at scale.
#
capjamesg[d]
I just like how Netlify's CI *just works*
#
hs0ucy
My server is right next to me ... in my basement
#
hs0ucy
with notes and bookmarks, a website grows quickly :S
#
capjamesg[d]
Love it hs0oucy. I think we have one or two people here who self host their sites.
#
capjamesg[d]
Yeah. I have lots of likes pages on my site now 😄
#
capjamesg[d]
[snarfed] I'll look into sentry. That solution did come to mind. Thanks!
hendursa1, kogepan and hendursaga joined the channel
#
[tantek]
capjamesg[d] the rel=me links summary thing looks cool! I would add icons for each of the services, e.g. in https://indieweb-search.jamesg.blog/results?query=tantek.com+social note that it doesn't look as quickly accessible as my Elsewhere sidebar on my homepage: https://tantek.com/
#
capjamesg[d]
Good catch! I’ll make that change soon.
#
capjamesg[d]
Is there an open icon set for this sort of thing [tantek]!
#
[tantek]
what are social media icons
#
Loqi
social media icon set is a collection of icons for various services, typically silos, that you can use on your homepage to link to your profiles on those sites https://indieweb.org/social_media_icons
#
capjamesg[d]
Anyway… the search Microsub thing was just a thought.
#
capjamesg[d]
Someone can do it if they want to. I think that’s what is so interesting about using microformats for marking up data: opportunities for using data become clearer / greater.
#
capjamesg[d]
I might subscribe to my own blog home page feed just for fun.
#
capjamesg[d]
And see what happens.
#
capjamesg[d]
[tantek] what other summaries would be useful? One person mentioned the idea of the phone book and how it is used to look up info about someone.
#
capjamesg[d]
Curious to dive deeper into that.
#
[tantek]
"Elsewhere" is pretty much that, a phone book for modern "phones" (communicators)
#
petermolnar
a thought/question: when one refers to algorithmical feeds does that imply and order change, a filtering, or both?
#
[tantek]
what is an algorithmic feed
#
Loqi
algorithmic feed (AKA algorithm-driven feed or just algorithm feed) is a more correct term for the "algorithmic timeline" lie, and an increasingly common feature on social media silos such as Instagram, Facebook, and Twitter, where they show only some posts from your followings, as well as show some posts only hours or days after they were posted, thus not in chronological order https://indieweb.org/algorithmic_feed
#
[tantek]
sounds like both 😄
akevinhuang2 joined the channel
#
[snarfed]
capjamesg the rel-me summary reminds me of https://snarfed.org/2014-03-10_lets-talk-userscript
#
[snarfed]
(...and sigh, looks like I need to update a link or two there, userscripts.org doesn't exist any more)
#
[tantek]
what is domain deaths
#
Loqi
domain-deaths are domain names that have been abandoned by their owner, and are either unresponsive, or vulnerable to being (or have been) taken over by bad actors, and thus any URLs to them should be unlinked and replaced with Internet Archive versions instead https://indieweb.org/domain-deaths
#
[tantek]
^ add it snarfed?
#
[snarfed]
sure! and oh wow we evidently have plenty of others to add there too
#
[tantek]
yes we do 😔
#
nekr0z
capjamesg[d]: just in case adding avatars to feeds as someone suggested earlier looks like a big effort, you can always leverage https://indieweb-glue.herokuapp.com/ ;)
#
capjamesg[d]
Wow. That is one long file.
#
capjamesg[d]
(Referring to snarfed’s comment)
#
capjamesg[d]
[snarfed] the search engine can render h cards in a standard way funnily enough.
#
Loqi
nekr0z has 3 karma in this channel over the last year (4 in all channels)
#
capjamesg[d]
I love these sorts of micro services.
#
capjamesg[d]
I have had to build a few of my own lately for IndieWeb stuff.
#
capjamesg[d]
The best way for me to do this would be in the crawl though. That would be much faster than issuing 10 network requests to retrieve the image URLs from another servers
#
nekr0z
capjamesg[d]: True; if you don't mind storing those in your index/database, that is the superior approach.
#
capjamesg[d]
I’ll probably just store the URL.
#
capjamesg[d]
I don’t have room for images haha.
#
capjamesg[d]
In fact the way things are looking I need to upgrade the server. The concern is memory though.
#
capjamesg[d]
Use your domain name 😄
#
[tantek]
haha the slug that one "aaron-parecki-totality"
#
capjamesg[d]
Code to fix that is in the next release.
#
[tantek]
that *won*
#
capjamesg[d]
I actually have no idea how that even got in the first 20 results.
#
capjamesg[d]
That is so odd.
#
[tantek]
slugs gonna slug it out
#
capjamesg[d]
I think it is an issue with the way that who is queries are treated. Good find aaronpk.
#
capjamesg[d]
I’m learning so much about search from all of you folks!
#
capjamesg[d]
/ challenging ways I am thinking
Seirdy joined the channel
#
[tantek]
capjamesg[d] — I think it'd be great to see how people tagged themselves in their homepage h-card
#
[tantek]
you could even those those tags/categories into links which then displayed lists of h-cards with those tags, so say, everyone who tagged themselves "runner" could find each other
#
capjamesg[d]
That is such a neat idea!
#
[tantek]
(that was a 55min delayed reply to "what other summaries would be useful? [about a person])
#
[tantek]
or say "coffee" 😄
#
capjamesg[d]
It plays into the social graph angle I have been thinking about. But in a whole new way.
#
[tantek]
and presto you have virtual opt in "groups" 😄
#
aaronpk
"circles" 😂
#
capjamesg[d]
I have already found some articles about coffee roasting of all things in the search engine. I found it so fun to make that discovery:
#
[snarfed]
heh h-cards and circles and graphs remind me of eg https://indiemap.org/social_graph.jpg
#
[tantek]
lol well that's definitely circles of h-cards 😂
#
capjamesg[d]
[snarfed] yeah. Your Kumu graph is so cool.
#
[snarfed]
thanks!
#
capjamesg[d]
[KevinMarks] about your idea…
#
capjamesg[d]
Reply contexts are interesting. I can imagine scrolling through the search with profile pics / names from h cards z
#
capjamesg[d]
And I’m going to remove the breadcrumbs. I like them but they take up too much room.
#
capjamesg[d]
Plus I don’t think they are really necessary.
edburns[d], gRegor, tetov-irc and akevinhuang joined the channel