#dev 2022-09-13

2022-09-13 UTC
Hgghn joined the channel
#
corlaez I love the idea of using CSS Hooks for microformats. In general I like empowering HTML (like the htmx library does)
#
corlaez I do see some issues... Parsing, while cheap is made over and over, by different servers that may be using different parsers that may produce different outputs
#
corlaez I think I will stick with the classes (CSS hooks) and generate as a post processing step other files such as an RSS atom xml etc.
#
corlaez That would open up my website to more consumers while still using the microformats to avoid repeating myself.
#
corlaez and I feel more in control of how others will interpret my data (unless they are parsing themselves)
#
[schmarty] corlaez: take a look at granary.io ! It can convert between html-with-microformats and other feed formats like atom
#
corlaez Nice! It does the trick for sure
#
corlaez Ideally, I would like to add it as a build step in my generation
#
corlaez does the site have an API? otherwise I will have to go manual
#
[schmarty] All transforms are URL-based, so that is the API
#
[schmarty] It's also open source and works as a Python library
#
corlaez oh nice, I will look to use it as an API. Nice project!
#
corlaez what is granary
#
Loqi granary is the social web translator https://indieweb.org/granary
#
corlaez What's the difference between projects and tools on the wiki? I see granary on tools (page I haven't seen before) but not in projects.
#
corlaez Would it make sense to add tools as a section in projects?
#
aaronpk what is projects?
#
Loqi There are many projects you can use to get your site on the IndieWeb, improve your IndieWeb support, or browse for inspiration for your own project; please note, some development ability and familiarity with command line tools will likely be required for you to use and improve these projects https://indieweb.org/projects
#
aaronpk i think "projects" are meant to be more full-featured things
#
aaronpk it is a little vague tho, and the projects page is definitely one of the less well defined lists on the wiki
#
corlaez schmarty, I didn't know you were mmg. I sent a PR to the indie web ring. just a heads up https://github.com/martymcguire/indiewebring.ws/pull/15.
#
Loqi [corlaez] #15 Add link to webring to join or learn more
#
corlaez https://github.com/martymcguire/indiewebring.ws/pull/15
#
Loqi [corlaez] #15 Add link to webring to join or learn more
geoffo, angelo, AramZS and gRegor joined the channel
#
[Murray] [tantek] Thanks, but I'm not sure that really provides a validation of why you shouldn't do it. I haven't yet read through the linked articles (I will) but it may better to actually provide a brief reason why this is considered a bad practice? It just reads as "don't do this", unless I'm missing something.
nathan[m] joined the channel
#
[Murray] Also, [schmarty], that post is precisely one of the arguments being used in this conversation for why this pattern would make sense. The justification is that the use of microformats in class selectors would actively help ensure their use across a wide codebase 🙂 (this also touches on Ben Myers semantic selectors: https://benmyers.dev/blog/semantic-selectors/ and common best practices found in methodologies like CUBECSS:
#
[Murray] https://cube.fyi/)
mambang[m], AramZ-S[m], Steve[m], zack[m] and Pyroxtheythem[m] joined the channel
#
[Murray] but tbc, I'm not personally advocating for this. Just struggling to come up with legitimate counterpoints as to why this isn't more widely used 😅
h4kor[m], mro, mpardalos[m], tetov-irc and barnaby joined the channel
#
[Murray] Okay, I had some time over lunch to read through the linked articles, but I'm not sure any of them are actually relevant to my question (or even that section of the FAQ) 🤔 The linked articles all seem to be discussing what an HTML class *is* but none of them really touch on CSS, beyond making a case for semantic CSS names. If anything, that suggests to me that using microformats as CSS hooks is _supported_ by the FAQ, as that woul
#
[Murray] help keep semantics intact. Other articles talk about ways to reduce class duplication, which again supports the idea that microformat classes could be multi-use: both as microformats, and as style hooks, to help reduce classes overall 🤷‍♂️
chenghiz_ and jacky joined the channel
#
corlaez [Murray] My take on the lack of CSS Hooks and html as a weak hypermedia in general is that the industry (pushed by big corporation) has shifted to a JS focus
#
corlaez The huge JS ecosystem, in all their different stages through time have always been to the detriment of plain old HTML and CSS
#
corlaez The huge JS ecosystem, in all their different stages through time have always been to the detriment of plain old HTML and CSS
#
corlaez The huge JS ecosystem, in all their different stages through time have always been to the detriment of plain old HTML and CSS
#
corlaez And the non semantic HTML aided by a ton of complex JS is what also is pushing some to say that "naturally" the backend should be "isomorphic" and be written in JS as well
#
corlaez this is a trap, a monoculture, and the best way to stay away from it is to empower HTML, reduce JS and generate the HTML in the backend lang of choice
#
corlaez Sorry for sending the send message multiple times, I am impatient and gut reaction is to slam the enter key when I don't see the message sent. This input is also not locking/clearing after I hit enter
#
corlaez In summary, CSS hooks is not popular not because it is a bad idea but because what is popular in the web today is not good.
#
corlaez what is hateoas
#
Loqi It looks like we don't have a page for "hateoas" yet. Would you like to create it? (Or just say "hateoas is ____", a sentence describing the term)
#
corlaez what is HATEOAS
#
Loqi It looks like we don't have a page for "HATEOAS" yet. Would you like to create it? (Or just say "HATEOAS is ____", a sentence describing the term)
AramZS and jacky joined the channel
#
[tantek]4 [Murray] your reasoning is exactly why a bunch of us originally *did* recommend using microformats class names for styling hooks. It makes sense in theory, and for some subset of use-cases, it mostly "worked". However, with enough longer term experience, it turned out that sites that did that ended up breaking their microformats when they did redesigns. So that assertion of "would help keep semantics intact" turned out to be false
#
[tantek]4 for one.
#
[Murray] @corlaez I was specifically talking about microformats from a conversation yesterday, but largely agree 🙂
#
[tantek]4 The only way you "help keep semantics intact" is by having a semantics consuming use-case that is actively in use either by users or by the developers e.g. a validator, linter, regression tester.
#
[tantek]4 If/when push comes to shove of semantics vs design (e.g. by using the same class names for both), an iteration of design "wins" and breaks the semantics. This has been the long term experience.
#
[Murray] That definitely makes sense. (Though in this instance, wouldn't the CSS be the consuming use-case?) It's also what I've been saying in the other chat 😄 I just can't back it up with anything concrete, so had hoped some people here might have written up experiences about that 🙂
#
[Murray] Thanks for the clarification though. At the least, I can go back and point to the FAQ and mention those thoughts. I think enough people put forward an "I have no data, but this feels like a bad idea" that the original poster is going to do something different for now
#
[schmarty] i feel like using semantic attributes and/or microformats for CSS selectors works when the person creating the CSS is doing so while testing the HTML semantics. it's a way of making it obvious when the HTML semantics break because the CSS breaks.
#
[tantek]4 yes [schmarty] is correct, it's useful for visualizing the semantics for testing purposes
#
[schmarty] but that kind of assumes some expert is in control of the CSS as a way of enforcing good markup
#
[tantek]4 it's only when you depend on them for the "primary design of the page" that you get into problems because it turns out the primary design of the page has other more important design constituencies
#
barnaby CSS isn’t really a consumer of microformats in that people writing CSS have no incentive to follow the mf2 spec
#
[schmarty] and that kind of control may not actually be common!
#
[tantek]4 [Murray] you're right that the FAQ only gives an "answer" and doesn't provide the reasoning behind it. That might be worth writing up in a separate page/post (which I thought we did somewhere, I just can't find it) and then summarizing/linking to it
#
[Murray] agreed there too @schmarty. In this instance, we were talking about personal sites, so I guess in that sense there is that level of control, but still requires you to keep track of it all!
#
[Murray] and yes, I'd say that would be useful [tantek], at least it would provide something concrete to point people at that feels "official" 🙂
#
[Murray] tantek++ at any rate, thanks for all the help schmarty++
#
Loqi schmarty has 13 karma in this channel over the last year (29 in all channels)
#
Loqi tantek has 21 karma in this channel over the last year (74 in all channels)
barnaby and jacky joined the channel
#
corlaez [tantek] Validation may be needed but I don't think it is a deal breaker
#
corlaez we could even have simple validation by coloring borders on microformats css.
#
corlaez we could even have simple validation by coloring borders on microformats css.
#
corlaez we could even have simple validation by coloring borders on microformats css
#
corlaez we could even have simple validation by coloring borders on microformats css.
#
corlaez oh jeesh. I need to get used to this
#
corlaez https://ibb.co/bPJCGsP
#
corlaez I guess what I am saying is, the need for validation/regression testing doesn't mean css hooks is bad. Arguably any approach will need some validation and regression when you tinker with the code
#
corlaez specially when tinkering with the code
jacky and AramZS joined the channel
#
aaronpk grr something is wrong with my char counting code
#
aaronpk it keeps truncating my tweets and posting a link to my site even when the text would all fit on twitter
cbzen and [naturestudy] joined the channel
#
[tantek]4 aaronpk you have no idea how much work I had to put into my own code to do that lol
#
[tantek]4 I *still* have a "preview the POSSE tweet" feature with a "characters remaining" count before publishing
#
[tantek]4 I'm interested in the content use-cases that you are finding errors with "even when the text would all fit on twitter"
#
aaronpk my last post for example
#
aaronpk i think it's because it has a link
#
aaronpk but i thought that was already accounted for
#
aaronpk and some of my posts with links are fine
#
aaronpk i think this is all done inside silopub
#
[snarfed] ah, silo.pub uses brevity, another kylewm library, for char counting. noticed that your silo.pub fork is on an older version of brevity than upstream
#
[snarfed] https://github.com/aaronpk/silo.pub/blob/master/requirements.txt#L2 vs https://github.com/kylewm/silo.pub/commit/46aece85f8918f56ed75f1e11b544c10f70a17fc
#
[snarfed] (from 4.5y ago though)
#
aaronpk let me see what i actually deployed
#
[snarfed] https://github.com/kylewm/brevity/#changes
#
aaronpk aha it looks like i am using brevity 0.2.11 that is missing a bunch of the char counting fixes
#
[snarfed] :thumbsup:
#
aaronpk now how do i update this heh
#
aaronpk i think this is the only python service i run
#
[snarfed] surgically, just on the server, activate the silo.pub virtualenv, then `pip install -U brevity`
#
[snarfed] more sustainably: update the brevity line in `requirements.txt`, then activate virtualenv and `pip install -r requirements.txt`
#
aaronpk hmm it looks like all my changes actually got merged upstream!
#
aaronpk well let's see if this fixes it
jacky joined the channel
#
aaronpk now i just need something to tweet
#
[tantek]4 aaronpk, what kind of API are you looking for for this functionality? perhaps I can see if I can improve my CASSIS functions to meet your requirements
#
aaronpk micropub-to-twitter
#
[tantek]4 oh wow ok much higher level than I've implemented
#
aaronpk including setting all the twitter properties like in-reply-to, and uploading photos and video
#
[tantek]4 I was asking about character counting for a POSSE destination 😂
#
aaronpk silopub just also happens to have a feature where if the text of your tweet is larger than the twitter limit, it will truncate it and add a link to your original post in the tweet
#
[tantek]4 got it. yeah my code to set in-reply-to is in Falcon PHP to call the Twitter API, and if I'm POSSEing photos or video to Twitter, then I skip my code and use Bridgy Publish
#
[tantek]4 on another topic, has anyone explored a nicknames cache use-case of recognizing "proper names" in text content and *suggesting* linking their first instance in the content to their canonical link from the nicknames cache, with the side-effect of replacing the proper name with a POSSE destination @-name in POSSE text?
#
[tantek]4 e.g. this tweet from voxpelli: https://twitter.com/voxpelli/status/1569721025046237192
#
@voxpelli This concept from The Verge also aligns with the @indiewebcamp philosophy of building and maintaining your presence online from your own site and only syndicating it elsewhere: https://indieweb.org/POSSE (twitter.com/_/status/1569721025046237192)
#
[tantek]4 "The Verge" really could (should?) have been @TheVerge in the tweet, and it seems that (semi-)automating that sort of thing would be helpful (rather than having to always personally dig up the @-name of an entity that you’re mentioning)
#
[tantek]4 or am I the only that would want this feature?
#
[tantek]4 (from a content authoring perspective)
#
[schmarty] i'm suspicious of automating those as i sometimes have trouble determining whether a suggestion is "right" with twitter's own @-suggestion UI
jacky joined the channel
#
[snarfed] [aaronpk] feel free to weigh in on https://github.com/snarfed/bridgy/issues/796
#
Loqi [snarfed] #796 micropub API for bridgy publish
#
[snarfed] [schmarty]++ agreed, but that may just mean twitter's @-autocomplete UI could be better
#
Loqi [schmarty] has 14 karma in this channel over the last year (30 in all channels)
#
aaronpk hah i just realized that silo.pub mentions your blog post as the inspiration [snarfed]
#
[snarfed] 😎
#
capjamesg Is there an mf2 h-event to iCalendar converter?
jonnybarnes joined the channel
#
capjamesg Secondly, how do calendars stay synced from an .ics file? Do they poll the file?
#
aaronpk yes
#
aaronpk https://github.com/gRegorLove/mf2-to-iCalendar
#
Loqi [gRegorLove] mf2-to-iCalendar: Convert microformats h-event to iCalendar
#
[schmarty] anyone doing ics + websub? 😏
geoffo and AramZS joined the channel
#
aaronpk oh darn no linkedin support for bridgy 😅
#
aaronpk oh darn, looks like probably not doablw https://github.com/snarfed/bridgy/issues/166
#
Loqi [snarfed] #166 add LinkedIn support
barnaby and [Jamie_Tanna] joined the channel
#
[Jamie_Tanna] See also https://rsvp-calendar.tanna.dev
#
Loqi Jamie Tanna
#
[Jamie_Tanna] (which as the name suggests is for a feed of RSVPs to iCalendar)
#
gRegor ooh neat, I'd missed https://authorship.rocks/
#
[tantek]4 [schmarty] re: "have trouble determining whether a suggestion is "right" with twitter's own @-suggestion UI" that's specifically why I noted it in the context of using a nicknames cache to solve this, because presumably you had to explicitly approve/add something to your own nicknames cache
#
[snarfed] on the plus side, personally, when I mention someone on Twitter, it's more often a new account I've never mentioned before, so Twitter's global namespace and autocomplete is easier than having to manually add that account to my local nickname cache first
#
[snarfed] (easier, but still not easy, since their UI is too minimal and it's hard to tell which suggestion is the right one)
#
gRegor I still usually think Twitter-first (for better or worse XD) so if I'm @-ing someone, I'd type their Twitter handle directly in the post
barnaby joined the channel
#
[tantek]4 gRegor, right, I do the same currently, however my goal with a nicknames cache is to type either their proper name or domain name into my text content and have my editor replace either of those with a personal-site linked proper name (or shortname / given name), with enough meta information so that the POSSE step automatically substitutes their @-handle local to the service
#
[tantek]4 there are two different UIs at work here:
#
[tantek]4 1 the UI to suggest / populate the nicknames cache itseslf. this is where using Twitter's suggested @-names may be helpful, and requires careful human / author review
tetov-irc and AramZS joined the channel
#
[tantek]4 2. the authoring UI that notices proper names / domains and uses the nicknames cache to provide semi-automatic replacements while writing (similar to what social media authoring UIs do, but using indieweb links/mentions rather than silo profiles)
jacky joined the channel