#dev 2022-01-05

2022-01-05 UTC
DerEinzigWahreMa and darkkirb joined the channel
#
[fluffy]
I leave that on because 97% of the time if I’m getting a random tweet from someone in those categories it’s them asking me to give them my username
#
[fluffy]
the other 3% is them demanding it
#
[tantek]
what a coincidence, I see similar behaviors
sarahd[d], jessealama and strugee joined the channel
#
@eay
↩️ I’m blogging on https://eay.cc since 2003. Mostly in German. Because I regularly post small status updates & links I rely on Good Ol’ WordPress with the simple non-RTE classic editor, which is fast for publishing/editing content. Also using Webmentions and JSON feeds.
(twitter.com/_/status/1478633461384724480)
jessealama, [snarfed], [fluffy]1, darkkirb, KartikPrabhu, petermolnar, Guest6, [tantek], tetov-irc, zerojames[d], hoenir, Murray[d], Jeremiah[d], shaunix[d], tracydurnell[d], MarkJR84[d], corenominal[d], wackycity[d], indieweb-irc-bri, grantcodes[d], Myst[d], daiyi[d], rattroupe[d], lanodan, oodani, edburns[d], Ramon[d], [jacky], push-f, nsh, Saphire, angelo_, nertzy, Abhas[m], reed, capjamesg, rhiaro, unrelentingtech, srushe, benji, kandr3s[m], sebbu, tomleo[m], samwilson, wagle, nekr0z, Zegnat, frittro, rommudoh[m], vilhalmer, npd[m], Charlotteracracs, benatkin, voxpelli, gerben, Kaja, willnorris, kloenk, PeterMolnar[m], LaBcasse[m], mambang[m], sknebel, [KevinMarks], klez, moose333, mikeputnam1, BinarySavior, eco_, Ruxton, stevestreza, chenghiz_, wrmilling_alt, GWG, bneil, lagash, alex11, superkuh, [manton], [tw2113_Slack_], vj--, saptaks, gillytech, strugee, [schmarty], IWSlackGateway1, aaronpk, [JampiDotNet]1, doubleloop[m], jamietanna[m], vikanezrimaya, jeremycherfas, hala-bala[m], rrix, micahrl[m], Seirdy, feoh, jjuran, joshproehl___, [JampiDotNet], doosboox, ben_thatmust, tetov-irc2, justIrresolute5, juanchipro[m], astralbijection[, sayanarijit[d], Nezteb[d] and [aciccarello] joined the channel
#
zerojames[d]
jessealama What are you hoping to achieve with adding structured markup to your site?
#
[tantek]
zerojames[d], re: your post, there's a differences among taxonomy, site information architecture ("sections of your site"), and post types. Yes they can overlap, but just because one feels awkward or unnecessary doesn't mean the others are.
#
zerojames[d]
Yes, there are 🙂
#
zerojames[d]
And I agree.
#
zerojames[d]
Likes and bookmarks are still their old structured selves and notes are marked up appropriately too. But, now there are no checkins, coffee posts, watch posts, reposts, and a few others that I have experimented with that I didn't use heavily.
#
[tantek]
just saying "launched a new taxonomy on my site" felt inaccurate
#
zerojames[d]
My overall thought is to use markup / post types that you really need rather than over-optimizing like I did.
samwilson joined the channel
#
[tantek]
when immediately followed up with "You are adding another section"
#
[tantek]
taxonomy != section
#
zerojames[d]
Ah, I see. I don't use "taxonomy" very often so I apologise for any semantic issues.
#
zerojames[d]
Thanks for sharing.
#
[tantek]
like you *can* makes sections for different terms in a taxonomy, but you can also have a taxonomy on your site *without* any new sections
#
zerojames[d]
Can you elaborate?
#
[tantek]
what is a taxonomy?
#
Loqi
taxonomy is an explicitly curated, often designed, set of categories or labels, often hierarchical as well, in contrast to the folksonomy of tags and hashtags https://indieweb.org/taxonomy
jackdaw[d] joined the channel
#
[tantek]
I can have a curated set of categories that I assign to my posts, without ever creating a new section, URL etc. on my site
#
zerojames[d]
That is true. That's the approach I am using now. Hashtags will correlate to new tag pages.
#
zerojames[d]
Thus, #watched will have a /tag/watched/ page where you can see all my watched.
#
zerojames[d]
*watches
#
[tantek]
sure, you can have dynamically created paths like that, I wouldn't call those explicit "sections" though
#
[tantek]
also tagging something as an #RSVP is not the same as a post having the "post type" of RSVP (simplified: determined by whether there's a p-rsvp property or not)
#
[tantek]
In short, your essay is somewhat interchangeably using "post type", "taxonomy" and "[site] section" when they mean very different specific things and are unfortunately easy to confuse
hala-bala[m] joined the channel
#
[tantek]
well, not necessarily specific, there's some fuzzy overlap which is what makes this even more confusing
#
[tantek]
e.g. "post type" is both the explicit notion of post type from the Post Type Discovery algorithm, and it is a broader sense of different kinds of posts, with different presentation, potentially only differentiated by an emoji or a hashtag
#
[tantek]
what is a post type
#
Loqi
Post or posts may refer to individual pieces of content published on an indieweb site such as notes, articles, & responses, or the act of creating the aforementioned content (present tense), or Posts about the IndieWeb https://indieweb.org/post_type
#
Loqi
[James] Overcomplicating post types (and committing to more simple ones)
rommudoh[m], mambang[m], DerEinzigWahreMa and diegov joined the channel
#
[tantek]
let me try again at a summary distinction (and why it's bad to conflate these things)
#
[tantek]
"post types" is https://indieweb.org/posts#Types_of_Posts, and ranges from different presentations of posts to explicit post types identified by markup (PTD algorithms), to sometimes "just" certain patterns in the content of posts (one or more specific emoji or hashtags)
#
[tantek]
"section on my site" is information architecture. how you choose to have what explicit sections on your site or not does not have to have anything to do with "post types", rather it should be about how what experience (UX) you want to present, what navigation paths, what areas of focus
#
[tantek]
there is nothing anywhere that says or implies that if you choose to distinguish a post type (of any kind) that you should or must have a section to do so
#
Murray[d]
what is a folksonomy
#
Loqi
tags or tagging refers to categorizing or labeling content, your own or others (tag-reply), with words, phrases, names, or other information, optionally linked to specific people, events, locations, such as the practice of tagging posts being about certain people (person-tag), like tagging people or other items where (area-tag) they're depicted in a photo https://indieweb.org/folksonomy
#
[tantek]
lastly a taxonomy https://indieweb.org/taxonomy is yet a different concept from any of those, if anything more abstract
#
[tantek]
in an abstract sense, the list of post types that we've been keeping track of on the wiki are a taxonomy of sorts, in that it is an explicit list of categories
#
[tantek]
and coming up with a taxonomy for what you're going to *name* the different sections of your site is also independently useful when doing information architecture. it does not mean that your site design is now "a taxonomy" though
#
[aciccarello]
I keep revisiting these Post kind vs architecture vs UX concepts anytime I add to my site. I still feel like I'm trying to wrap my head around them. Thanks for expanding on the differences.
nekr0z joined the channel
#
[tantek]
also creating a new name or a new type or a new category is not a "new taxonomy"
#
[tantek]
it's adding something to an existing taxonomy
#
[tantek]
zerojames[d], apologies for using your blog post as an example in this regard, it did seem like a good case study for expanding on some of these sources of confusion
binyamin[m], npd[m], [jeremycherfas] and [chrisaldrich] joined the channel
#
[tantek]
zerojames[d], re: "expand my notes field to do things like automatically add RSVP markup" <- exactly! My "RSVPs" are literally just "special" notes that my code recognizes from the first few words of the note, and then automatically adds the RSVP markup, and alters the presentation accordingly. I still write them as "notes".
Abhas[m] joined the channel
#
[tantek]
There's lots of ways of doing different types/kinds of posts, and there doesn't have to be a 1:1 match of creation UX vs display UX
EvanBoehs[m] joined the channel
#
[tantek]
you could literally create every post as a "note", then have your code analyze it and decide to display it differently based on the contents of the note
#
[tantek]
I actually think that's a better creation UX. No need to burden the author with "pick a post type" at all. Just start typing or uploading a photo or whatever
#
[tantek]
a lot of us in the IndieWeb community initially got distracted by the "pick a post type" UI initially because frankly that's what Tumblr did and that seemed to be popular.
kimberlyhirsh[d] and Lohn joined the channel
#
[aciccarello]
Personally I want to distinguish photo posts vs notes with posts which I think doesn't fit with post type discovery. Like I have photos I want to share which I want to distinguish from "here's a note with some related image". Does anyone else do that?
#
[aciccarello]
Maybe I'm too focused on making my photos page look nice 😂
#
[tantek]
photos are one of the biggest uses for post type discovery!
#
[tantek]
why do you think distinguishing photos vs notes "doesn't fit with post type discovery"
reed and micahrl[m] joined the channel
#
aaronpk
one thing i've noticed myself is wanting to distinguish my notes with screenshots from my photo posts
#
[aciccarello]
Yeah, that's one scenario I've thought about.
#
[aciccarello]
tantek if they both have an image and a text body how would they be distinguished?
PeterMolnar[m] and kandr3s[m] joined the channel
#
[tantek]
aciccarello, FWIW I personally disthtinguish "photo" vs note with related image by what comes fist in the content. photo (or photos) at the start? that's a photo (or multiphoto) post. otherwise note with whatever else
#
[tantek]
first* in the content
#
[aciccarello]
Ah, that's interesting. I'd have to think about how that would work with my site since I store featured images in the frontmatter of a markdown file.
#
[aciccarello]
I also don't have micropub setup yet
#
[aciccarello]
I probably should push past some decision paralysis and experiment more. I'd have to be okay with possibly reorganizing things later 🤔
#
Murray[d]
if you are already using frontmatter, couldn't you just override the post type there? So if it has an image, it's a photo, unless the frontmatter says otherwise?
#
zerojames[d]
[tantek] I have just deleted the post and might rewrite it to use the "post type" syntax instead.
#
[tantek]
oh no! I was worried I might give that impression. I think it only needed some light editing zerojames[d]! I got what you were trying to say
#
[tantek]
tbh this is the kind of thing that works much better in an interactive a/v chat.
#
[tantek]
when is HWC?
#
[manton]
I thought it was a good post and wouldn’t take it completely down either. You could edit it, or even just add an “Update:” at the end with a summary of some of this conversation.
Eddy04[d] and jamietanna joined the channel
#
jamietanna
zerojames[d] / capjamesg FYI https://jamesg.blog/category/post (linked from your posts) is 404ing, and https://jamesg.blog/2022/01/04/simple-taxonomies/ doesn't appear to be deleted? (as you'd mentioned you had deleted it above)
caro401_she_her_ and KartikPrabhu joined the channel
#
[jacky]
in the (odd) spirit of DIY, I want my site's search to be in-built
#
[jacky]
but since I separate my presentation site from the parts of it that run (like Micropub and Webmention), I'm tempted to have it be _yet another thing_ that listens to changes from my WebSub server to know when to add something to the index and also maybe get more information semantically from Micropub
#
[jacky]
files this away as an experiment for a rainy day
#
zerojames[d]
Ah that is a downside of my current GitHub Actions setup.
#
zerojames[d]
I need to remember to manually purge deleted pages, which I only do every so often.
#
zerojames[d]
I wonder if I can update my action to do this.
#
zerojames[d]
[jacky] Good idea 😉
#
zerojames[d]
I wrote a basic crawler to index my site last year.
#
[jacky]
it'd be mainly for my site tbh
#
[jacky]
oh nice
#
zerojames[d]
(Before indieweb search, that is)
#
zerojames[d]
And I loved having my own search feature!
#
zerojames[d]
I offered image search and such because I could. The sample size was small enough 🙂
#
zerojames[d]
Let me know if you build anything. Happy to answer questions to the extent I can!
#
zerojames[d]
The real bonus of running your own engine aside from being able to crawl on demand is that you can get creative with the search results if you want.
#
zerojames[d]
I had my search feature as another thing in my site, hosted on a subdomain.
aspen[d] joined the channel
#
[jacky]
I might make into a custom page on my site
#
[jacky]
my hope was to be able to provide a local way to find things
#
[jacky]
it's either that or me implementing some sort of "other things" based on how it's linked across my site (more feasible for me)
#
zerojames[d]
How do you store data on your site?
#
zerojames[d]
If you store most things in a DB (or at least the things you want to search), you don’t really have to write a crawler or anything like that.
#
zerojames[d]
I have found a lot of DB engines have powerful full text search built in.
#
[jacky]
it's mixed
#
[jacky]
some "indexable" content like slugs are stored in a database
#
[jacky]
and category + channel associations
#
[jacky]
but the content itself is all on disk
#
[jacky]
like flat file
#
[schmarty]
the benefits of setting up search as its own service is you can almost focus on just the bits you need/want to make a good search
#
[jacky]
I might have to have something run analysis on each post, build an index and maintain it over time for something like this (tbh the index could be _another_ files called `referenced_by.txt` that just has URLs of things that have linked to this entry and `referencing.txt` being links that go out)
#
[schmarty]
the risks are of course that "almost" - overhead for running a separate service and coordination for getting the actual data _in_
#
[jacky]
[schmarty] yup
#
[snarfed]
the other interesting option is client side, ie you ship your entire index with your site and search it client side, in the browser. requires JS, but surprisingly small and feasible for many individual sites
#
[snarfed]
and zero latency enables some UX features that you can't realistically do with server side search that incurs network round trip(s) for each new set of results or data
#
[jacky]
true! thanks for reminding me of that; might make that a optional function
#
[aciccarello]
Yeah, that's one of the easier options if you already have a static site. A project I worked on used lunr
#
[aciccarello]
But it seems like a lot of docs sites use angolia now
#
[snarfed]
[jacky] maintaining two separate search stacks for client and server side sounds...heavy, and likely to skew
#
[jacky]
oh I'm hoping that the index I use for the client can be used for the server as well
#
[jacky]
like a singular index
#
[snarfed]
ah sure, the crawl and indexing would be shared, but not the search code. unless your web server is on Node and the JS is shared?
#
[jacky]
ah no it wouldn't be
#
[jacky]
might have to think on that some more
#
[snarfed]
probably start with just one
#
[jacky]
yeah hah
#
[jacky]
def the server side
#
[jacky]
it'll be v0: duckduckgo backed, v1: server-backed v2: ???
#
[snarfed]
the other benefit of client side is that it's less admin tax. have to regularly rebuild the index either way, but you don't have to build or run a search server
#
zerojames[d]
Indeed!
#
zerojames[d]
Running a search server does take admin work, even if you have a relatively simple index.
#
zerojames[d]
I used to have a crontab job set up that crawled my site every day. The crawl took a few mins and after that the whole index was up to date.
#
zerojames[d]
I haven’t played that much with client side search
#
zerojames[d]
But I do know you can make nice search as you type interfaces with a good back end.
#
zerojames[d]
Just like how Algolia does it.
#
zerojames[d]
[schmarty] +1 re; your comment about building what you need.
#
zerojames[d]
If your data is in flat files, you can write a crawler that just crawls the files rather than web pages if you want.
#
zerojames[d]
That would reduce crawling time significantly and keep things simple on your end.
#
[jacky]
heh I could even make it incremental by checking the modification times on files that represent the index
#
zerojames[d]
Yep!
#
zerojames[d]
How many pages do you have on your site?
#
[jacky]
the live one has like 6 lol
#
[jacky]
my _older_ experiment had around ~3200
#
[jacky]
mainly notes or photos I've POSSE'd to Twitter or PESSO'd from Instagram
#
zerojames[d]
[manton] [tantek] good points! Thanks! I’ll keep the post up and maybe write a follow up re: taxonomies vs. post types.
marksuth[d] and [James_Van_Dyne] joined the channel
#
[James_Van_Dyne]
zerojames: I’ve put in a couple of PRs for your indieweb-utils library. I’ve got a couple more (small) ones that I’d like to make that build off my current PRs (introduce black/flake8/mypy). Not sure if you got notifications about the open PRs? I can’t request a review from you in GH for some reason. If there’s a different way you’d like to collaborate on the project, let me know 🙂
#
zerojames[d]
I did indeed!
#
zerojames[d]
I shall take a look at them in the next few days. I greatly appreciate your contributions!
#
[James_Van_Dyne]
Awesome 🙂
capablecable[d], Christian_Olivie, [JampiDotNet], [tantek], [schmarty], justIrresolute, RIZY101[d] and tetov-irc joined the channel