#dev 2022-02-07

2022-02-07 UTC
dtw, Seirdy and Seirdy0 joined the channel
#
jacky
random: I was thinking again about owning my gameplays (and achievements and stuff)
#
jacky
(pingign [schmarty] b/c I think you're interested in this lightly)
#
jacky
but if one were to compile their achievments, would they want it to be on one page or like a new page per achievement?
#
jacky
I feel like the latter would get _very_ noisy
[schmarty] joined the channel
#
[schmarty]
Ooh hello yes
#
[schmarty]
a single page seems reasonable. Individual achievements could have fragment ids
#
jacky
excellent
#
jacky
was just updating the spec I had for it
[fluffy]1, KartikPrabhu, blink9, Seirdy, balupton[d] and tj1 joined the channel
#
@frankmeeuwsen
↩️ Maak van je eigen site een identiteitsprovider voor sites die dat ondersteunen. Zoals inloggen met Twitter , Facebook of Google. Maar dan met je eigen domein. Zie https://diggingthedigital.com/indieauth-wordpress/
(twitter.com/_/status/1490584315524268036)
tetov-irc joined the channel
#
[tantek]
The replies here 😂 (some of which are vaguely indieweb-related, toolchain stuff etc.) https://twitter.com/toddmorey/status/1489613725636173825
baracurda and baracurda_ joined the channel
#
kinduff
[tantek]: this one is my favourite, https://briefs.video/videos/how-do-i-draw-a-line/ saw it posted there too
#
kinduff
This guy Heydon Pickering is pretty awesome, has a video on the Indie web too
#
[tantek]
heydon++ yes he's got quite the sense of humor about this stuff
#
Loqi
heydon has 1 karma over the last year
#
capjamesg[d]
What is the best way to add HTML into a contenteditable field?
#
Loqi
It looks like we don't have a page for "best way to add HTML into a contenteditable field" yet. Would you like to create it? (Or just say "best way to add HTML into a contenteditable field is ____", a sentence describing the term)
#
capjamesg[d]
Oh Loqi, I thought you would know that 😂 just kidding.
#
capjamesg[d]
The use is for automatic highlighting when @ mentions are used in a text field.
#
capjamesg[d]
The field is contenteditable vs. textbox for this reason.
#
capjamesg[d]
I built something that works but for some reason the cursor gets sent back to the start of the field.
#
capjamesg[d]
So "@tantek Great blog post about websub!" would become "Great blog post about websub! @tantek" if I just kept typing.
#
kinduff
omg is heydon here?
#
kinduff
screams like a groupie
mbrozirkb__ joined the channel
#
kinduff
Webmentions via n8n with a webhook are working like a charm 😍
#
@kinduff
@rem Hey rem, wondering if the RSS feed for http://webmention.app needs to be formatted in a certain way, should it include source and target too?
(twitter.com/_/status/1490682522317033477)
[manton] joined the channel
#
[manton]
[aaronpk] Another side effect of including HTML on events.indieweb.org that could be improved… Kimberly’s RSVP has an h-card with “display: none” to hide on her blog, but the image shows up in the RSVP.
#
[manton]
Wonder if maybe all images should just be stripped out. I’m sure it won’t be perfect but seems like usually they don’t need to be shown.
jacky and nertzy joined the channel
#
aaronpk
hmm, I think this is not what I would expect the markup to look like tho...
#
aaronpk
It's pretty unusual to put an h-card *inside* the e-content
#
Loqi
[Kimberly Hirsh] I Kimberly Hirsh am RSVPing yes to Micro Camp 2022 .
#
jacky
that is
#
jacky
I could see it being used to tag other people in it
#
jacky
but for an *-author, I can see that being weird
#
[manton]
Yeah, usually the h-card would be somewhere else on the page. But I don’t think it’s necessarily wrong, either. I’ve used this kind of markup myself.
#
aaronpk
Yeah I mean specifically the author h-card
#
aaronpk
I do think it's "wrong" as demonstrated by the fact that you used display: none
#
aaronpk
of your goal is to hide it from humans reading it, you should also hide it from parsers, which you do by not putting it in the e-content
#
[manton]
Hmm. In this case, there is no way to put it somewhere else except e-content. You could say that’s a limitation in Micro.blog, but I can imagine other systems working like that too.
#
aaronpk
*ahem* Wordpress
#
aaronpk
why is there an e-content class if there isn't already an author?
#
[manton]
Looks like this theme doesn’t have any author markup.
#
[manton]
So I’m sure that’s why they put the h-card in the e-content… Basically a choice between modifying the theme or just pasting in some extra HTML in the post.
#
aaronpk
there's no author markup on the home page either?
#
[manton]
Doesn’t look like it.
#
jacky
side-note: these are interesting things to collect for a 'mf2 lint/doctor' tool
#
[manton]
I did the same thing a couple years ago here, probably for the same reasons: https://pin13.net/mf2/?url=https://www.manton.org/2020/06/08/ill-be-attending.html
#
Loqi
[Manton Reece] I’ll be attending IndieWebCamp West later this month. It’s online-only, centered in the west coast timezone, but anyone around the world is welcome. Need to think about IndieWeb-related goals for Micro.blog that week.
#
[manton]
I totally get that this is a little hacky, but I think it should be allowed, personally.
#
aaronpk
If there was an h-card on the home page, then the post could have a link like <a href="/" class="u-author">Kimberly</a> and the receiver can go fetch the full h-card from there and you can also show that link because it makes sense
#
[manton]
Yeah, that’s better.
#
aaronpk
thinking about this generically (since one-off examples often make sense but you forget the implications of something like that in the larger context), this seems very similar to the issue of putting photos inside the content with u-photo as well
#
sknebel
it's always a mess for consumers to clean up
#
aaronpk
its almost like what you want the parser/consumer to do is: if an HTML element inside an e-* contains other microformats properties that result in properties being added next to the e-* property, then you want to remove those elements from the HTML of the e-*
#
aaronpk
but I'm also not sure if that would break a bunch of other things too
#
[manton]
The nice thing about including photos inside e-content is that the HTML can be used in a lot of different contexts (like feeds) without requiring any special behavior to look right.
#
sknebel
it would break on your example with a link on the name
#
sknebel
or you could use the html you use for display instead of just the e-content in feeds
#
sknebel
you can even *store* that if you want to access it statically
#
[manton]
Just speaking for myself, there is no difference between e-content and the HTML I use for display.
baracurda_ joined the channel
#
aaronpk
Lol sknebel you're right, this is why test cases are useful :-D
#
aaronpk
Now I'm confused [manton], images *are* allowed in HTML e-content right now, which is causing the problem with the RSVP ;-)
#
Loqi
[aaronpk] #16 consider not including img alt text as part of surrounding text properties
#
[manton]
Okay, maybe I’m getting derailed with the u-photo discussion. Let’s take one issue at a time. 🙂
#
GWG
Who has spoken the word of power?
#
Loqi
[aaronpk] #52 Remove images from posts containing a photo
Ramon[d] joined the channel
#
aaronpk
they are similar because both are an example of adding additional microformats in the e-content that you then want removed from the e-content
#
GWG
Oh...more theme WordPress weirdness
#
aaronpk
GWG: just the same situation manton is in where people can add arbitrary html including microformats into the content box and the theme already has its own microformats outside of it that the user can't modify
#
GWG
aaronpk: I came up with a hacky solution for that...but the design of WordPress makes consistency problematic
#
GWG
I only insert e-content around the origin content..not anything added dynamically.
#
GWG
original.
jacky joined the channel
#
[manton]
Maybe let’s take a step back. Assuming the user can’t modify anything except what’s in e-content, is there a best practice for including h-card info?
#
jacky
I've always assumed it's been as a sibling to e-content/p-summary stuff
#
aaronpk
back to the specific problem, I'm actually confused why the event site is showing that at all because I thought my quick fix the other day was to show only the text content of comments
#
aaronpk
Oh lol other way around
#
sknebel
its ignoring the css hiding the text, of course
#
aaronpk
there was a problem with the text content because of the new line handling so I show the HTML now
#
[manton]
For events, it seems ideal to allow HTML but strip out images and do other sanitization. That way you can have links but it doesn’t look too cluttered with other stuff.
#
aaronpk
removing all images would break it for blog posts with inline images
#
sknebel
and images are not the problem here
#
sknebel
well, not only
#
sknebel
filtering the image would still leave the text from the h-card
#
aaronpk
I suppose you could argue that specifically for showing comments on a post or event, you would just always remove images in the content. That'd also break reaction gif posts tho
#
sknebel
(re what's the workaround: a link to a fuller h-card on something that's part of the text or an empty link, that at least doesn't show up visibly)
#
aaronpk
yea that was my suggestion earlier
#
aaronpk
Also note that link doesn't *have* to be to the home page, it can be to any page with an h-card
#
[manton]
So, I looked into what Micro.blog is doing here… And it’s ignoring “display: none” too. But apparently I’m styling the h-card photo so it’s smaller/inline: https://micro.blog/kimberlyhirsh/12444267
#
sknebel
because you style all images in h-cards
#
Loqi
[kimberlyhirsh] I Kimberly Hirsh am RSVPing yes to Micro Camp 2022.
#
aaronpk
I'm on mobile right now, what css is that? Is it going to make every image small? Or just h-cards?
#
sknebel
.post_text .h-card img {
#
sknebel
height: 32px;
#
sknebel
border-radius: 16px;
#
[manton]
That’s it. I must’ve noticed this problem and that was my “fix”.
#
aaronpk
I think I wouldn't be opposed to adding that too
#
sknebel
thus showing people its ok to do, because it doesnt break where they see it :/
#
GWG
[manton]: You also special cased my svgs, didn't you?
#
aaronpk
But it's not entirely wrong, just not ideal
#
[manton]
I think [aaronpk]’s point about blog post comments is good too… Those should allow at least some images, similar to RSVPs, but you don’t want the text taking over the page with all sorts of styling or big images.
#
[manton]
@GWG I do something similar for WordPress emoji PNG/SVG, maybe that’s what you’re thinking?
#
[manton]
(Where WordPress uses images instead of real emoji.)
#
[KevinMarks]
the bridgy html feeds are fun like that as there's no constraint on image size so you get huge images
#
[KevinMarks]
er, granary not bridgy
#
aaronpk
Side note: if the response has a name then only the name is shown and no content
#
aaronpk
but that concern applies to comments with lots of text/images but no name
#
sknebel
such images that are intended to be small are especially a problem
#
sknebel
inline photos, ok, you can limit them so they fit the layout and its mostly fine,
#
sknebel
but here figuring out that its actually supposed to be tiny is tricky
#
sknebel
unless you go very much in the weeds of whitelisting parts of CSS, *shudder*
#
aaronpk
Yeah that's the same WordPress emoji Robles
#
aaronpk
Problem
#
[manton]
Yeah, I have some CSS for WordPress looking for “img.emoji” and “img.wp-smiley” and probably others. And when looking for specific HTML, I actually have a list of filenames like “1f3b5.png” for books.
#
sknebel
(is that list somewhere public? sounds like a useful shared resource)
#
[manton]
Heh, very. 🙂 I think it’s a losing battle for WordPress since emoji are added every year, but… Shrug.
#
[manton]
No, but I’m happy to publish it. It’s not all emoji, just the couple dozen that I currently care about.
#
[manton]
There is probably a way to generalize it to all WordPress emoji.
#
aaronpk
I kind of wish there was an html element or attribute to indicate an image is part of inline text instead of decorative/large
#
[manton]
That would be really nice.
#
aaronpk
Isn't this also an accessibility concern? Any work done there to look at?
#
[KevinMarks]
interesting - like a sparkline attribute?
#
[manton]
I went to test this in WordPress and it looks like they’re no longer replacing emoji with images by default! Nice.
#
[manton]
Definitely an accessibility problem so I was wondering if WordPress already put something suitable in “alt”. But now don’t have a good reference for what WordPress used to do.
#
aaronpk
twitter of course doesn't have any useful markup around their img emojis
#
aaronpk
hm, should this h-card example use aria-hidden="true"?
petermolnar joined the channel
#
sknebel
display: none should be enough
sarahd[d], daiyi[d], Asaf_Agranat[d], Jeremiah[d], rattroupe[d], aspenmayer[d], sayanarijit[d], Seb[d], Murray[d], shaunix[d] and Zegnat[d] joined the channel
#
sknebel
aria-hidden="true" should not be added when:
#
sknebel
The element or the element's ancestor is hidden with display: none
#
sknebel
The element or the element's ancestor is hidden with visibility: hidden
#
sknebel
The HTML hidden attribute is present
corenominal[d], aaronpk[d], hans63us[d], capjamesg[d], Osvik[d], hoenir, mossymaker[d], Myst[d], Ramon[d], balupton[d], P1000[d], wackycity[d], marksuth[d], dovedozen[d], Christian_Olivie, MarkJR84[d] and barnaby joined the channel
#
aaronpk
yeah it seems like hiding stuff in the e-content with display: none and then being surprised when it shows up in something that consumes the e-content is the real problem here
#
aaronpk
so i think i have 2 options for the event site. either I add similar css as micro.blog to make that image appear small and inline, or i allow the "display: none" css
ben_thatmustbeme and cygnoir[d] joined the channel
#
GWG
You can turn off the JavaScript that replaces emojis with images I believe...
jacky joined the channel
#
[manton]
[aaronpk] The quick fix is definitely the CSS… I think the problem I ran into with “display: none” is being smart about allowing it _without_ also allowing everything else that can go in a “style” attribute.
#
aaronpk
yeah i'm reading the htmlpurifier docs now to figure that out
#
[manton]
Cool. If that already supports it, that would be great. I don’t think the library I use (“sanitize” Ruby gem) didn’t, although I should check again.
#
aaronpk
i guess the other option is i could just remove images from responses for the event site
#
aaronpk
that would turn it into "I Kimberly Hirsh am RSVPing yes to Micro Camp 2022"
#
[manton]
Oh, this gem actually does support it. I just don’t have “display” listed on my allowed CSS properties. Hmm.
#
[manton]
Removing images seems like a reasonable solution to me for RSVPs. I can’t imagine that images are used very often in RSVPs.
#
aaronpk
this isn't just for RSVPs though, this is for all comments on an event
nertzy joined the channel
#
aaronpk
tho you do bring up another point which is if a response *is* an rsvp, then i could remove images
#
[manton]
For completeness, here’s what I currently allow in HTML when showing a post in the Micro.blog timeline. I’m sure I’m missing some things… I add things that are safe when I notice something that doesn’t look right. https://gist.github.com/manton/48a5cd6148438d48c81e47e438cfc125
#
[manton]
Yeah, good point, maybe RSVPs have a different behavior. I forgot that events support non-RSVP comments.
#
aaronpk
hm there does seem to be a config to allow some css attributes
#
aaronpk
i'm a bit more nervous about that, i think i might just not show images at all
#
aaronpk
ok yeah, going to go with that for now since it's less of a drastic change. photo responses are also still supported too since those are treated entirely differently and will pull photos into a nice photo grid
#
aaronpk
oh, and that also more or less reverts it back to the previous state it was in a few days ago when only the plaintext content was used
#
aaronpk
that was a lot of faff for what ended up being just adding 8 characters to the code
#
aaronpk
alright that's deployed
#
[manton]
Heh. Cool!
edburns[d] joined the channel
#
[manton]
Looks good. 👍
#
barnaby
many of the best changes end up being hours of work for a few characters/lines changed in my experience
#
Loqi
lesscode has 1 karma over the last year
#
aaronpk
lesscode++
#
sknebel
(local record is something like 6-8 person-weeks for "not a software problem, here is proof, now go away and bug your CPU vendor instead")
#
GWG
I need to reread this when I take a long break
sayanarijit[d], Christian_Olivie, corenominal[d], hans63us[d], Myst[d], marksuth[d], MarkJR84[d], Zegnat[d], hoenir, capjamesg[d], aspenmayer[d], petermolnar, Jeremiah[d], Murray[d], shaunix[d], sarahd[d], Seb[d], Osvik[d], Asaf_Agranat[d], aaronpk[d], rattroupe[d], cygnoir[d], mossymaker[d], wackycity[d], dovedozen[d], edburns[d], balupton[d], Ramon[d] and daiyi[d] joined the channel
#
[snarfed]
lesscode++
#
Loqi
lesscode has 2 karma over the last year
#
[snarfed]
aaronpk while you're here, low priority, lmk if you want me to file a bug for https://chat.indieweb.org/dev/2022-02-06#t1644170249510000
#
aaronpk
i think that is correctly saying it is not a valid URL
#
aaronpk
IIRC · is not a valid character in URLs
#
[snarfed]
I got that from copying and pasting the URL in my browser into the webmention form at the bottom of https://blog.reiterate.app/software/2022/02/06/comments-enabled/
#
[snarfed]
looks like that form posts directly to https://webmention.io/reiterate.app/webmention , no JS
#
aaronpk
hmm, and that input is type=url
#
[snarfed]
so, not sure where in the chain we should try to improve/fix this
#
aaronpk
that means the browser is allowing those chars cause it complains when it otherwise doesn't look like a url
#
aaronpk
that's confusing
#
[snarfed]
i'm 95% sure the wm.io failure was interactive, and not the automatic one sent by my site, but I can confirm if you want
#
aaronpk
i think that is its URL validator giving that error
#
aaronpk
before it even attempts to fetch the URL
#
[snarfed]
yeah confirmed
#
capjamesg[d]
I have been beefing up my Microsub post interface. I can now auto link to person tags and get a list of recent hashtags I have used that match what I am trumping.
#
aaronpk
wow nice!
#
capjamesg[d]
There is now a visual indicator for uploaded images (no more HTML in the client input field for images!). And I can use emojis with the : syntax.
#
capjamesg[d]
writing all of this in raw JS has been quite tedious and done over quite a long time.
#
capjamesg[d]
The person tags come from the proposed micropub contact extension my endpoint supports. And the hashtags come from my micropub endpoint too.
#
[snarfed]
capjamesg++
#
Loqi
capjamesg has 26 karma in this channel over the last year (72 in all channels)
[Aaron_Klemm] joined the channel
#
capjamesg[d]
And embedded rely contexts
#
capjamesg[d]
[James_Van_Dyne] you are in my contacts list 🙂
#
capjamesg[d]
I don't support the tooltip functionality because I haven't written anything close to a tooltip just yet. But I like the idea.
#
capjamesg[d]
There's the hashtag function right now ^
jacky joined the channel
#
[tantek]
what is a photo comment?
#
Loqi
photo comment is what a photo reply looks like in the context of the original post that it is in-reply-to https://indieweb.org/photo_comment
#
[tantek]
I still think photos make sense inside e-content because they're literally part (if not the primary aspect) of the visible content being presented
#
[tantek]
we should expect that there will be lots more use-cases for additional properties or nested microformats inside e-content
#
Loqi
I agree
#
jacky
heh Loqi
#
aaronpk
there's no concern with photos inside e-content if they are part of the content, it's only a problem when they are *also* marked up with u-photo
#
aaronpk
it's effectively duplicating the photo since it then appears both in the content and outside the content
#
jacky
feels like that should be mentioned as a warning of sorts on /photo
#
jacky
or on the mf wiki
[jgmac1106] joined the channel
#
aaronpk
here's another way of thinking about it... the point of adding microformats markup is to *help* consumers of that data do something with it. if adding markup is actively making it more difficult for the consumer then don't add the markup. in other words, having a photo inside the content is fine with no other markup of the photo. but if you want to help consumers do something smart with that photo (show
#
aaronpk
it as a photo post, etc), then you can *add* the u-photo class to *help* them
#
barnaby
that reminds me, I need to set the default mf class for uploaded photos in my posts to u-representative, as at the moment all of my images are within the e-content
#
aaronpk
do you mean u-featured?
#
barnaby
oops yes you’re right. it’s been a while :D
#
barnaby
heh, the spec even uses both terms: “u-featured - a representative photo”
#
[jgmac1106]
and barnaby build some tool that will automatically take a u-featured and turn it into OGP kruff for me
#
[tantek]
aaronpk, adding u-photo *is* helping consumers of that data do something [different] with it, it's indicating that it's a photo post and should be presented with the photo as the focus of the post.
#
barnaby
I guess I should do a mass find/replace of u-photo with u-featured, too
#
[tantek]
barnaby huh? only if you intend to change the kind of post
#
aaronpk
[tantek]: right, in which case if you're going to go to the trouble of adding that, then the photo should not also be in the content
#
[tantek]
except it's not "also" in the content, it is the content so it makes sense to be inside e-content from a publishing perspective
#
aaronpk
has a sore spot from dealing with removing images from content that are also in the photo property when building a reader
#
[tantek]
yeah that's the challenge worth documenting for sure
#
aaronpk
my point is that doubling up the photo is actively making it harder for consumers, which is kind of defeating the purpose of microformats
#
aaronpk
you're better off *not* putting the u-photo on it if your photo is inside the content
#
[tantek]
it's not doubling-up, the photo is only in the post markup *once*
#
aaronpk
it's in the resulting mf2 twice
#
aaronpk
and that's what a mf2 consumer sees
#
[tantek]
lots of things get parsed out multiple times in mf2
#
[tantek]
that's for thoroughness, not convenience
#
aaronpk
and in all the other cases it's useful and helpful
#
[tantek]
from the publisher's perspective, it's not "doubled" so it's wrong to call it that
#
barnaby
heh, I remember discussing this a while ago. interesting to see that it’s still a talking point. In theory I like the idea of e-content being self-contained, so that consumers can adapt to arbitrary post types by just using e-content as a fallback, and then having more advanced behaviours for additional properties which they support (e.g. recognising that u-photo is duplicated in e-content, and making sure that it’s only
#
aaronpk
then there's something fundamentally wrong with the microformats parsing rules
#
barnaby
displayed once)
#
[tantek]
barnaby, that's the idea
#
aaronpk
if the publisher doesn't think they are creating duplicate data but the consumer sees duplicate data, then there's something wrong
#
[tantek]
it's not duplicate data, it's different filters of the same data
#
[tantek]
in this case, generic consuming apps can go with e-content and be done
#
[tantek]
if you have a photo-specific app you *want* to only show photos posts. I believe there are even views in one of the readers that does this
#
[tantek]
both are useful
#
jacky
^ but even in that case, if the photo's doubly tagged, does the photo-specific app _now_ have to scan the HTML for values existing in `photo` and prune them?
#
[tantek]
this is a problem specific to a particular kind of reader, not a problem with mf2 parsing
#
aaronpk
right jacky, a consumer sees the "photo" property and is like cool this is a photo post. and then if it also decides to display the content, now the photo appears twice
#
aaronpk
if you're saying the content property can be treated *only* as a fallback, then that's useful
#
[tantek]
same thing with video, audio etc.
#
aaronpk
but if the content property is expected to have something useful when there is also a photo property, there's the problem
#
jacky
yeah this actually makes me reconsider how I'd mark up my photos then
#
aaronpk
if you want to have fallback behavior for consumers that don't understand a specific property, then that fallback should be able to be ignored when consuming the new property
#
[tantek]
no it's not just "fallback"
#
[tantek]
and there can be useful things in content that are beyond the photo that you don't display immediately
#
[tantek]
e.g. a photo grid display reader view only displays the photos
#
aaronpk
that's my point, there *shouldnt* be if you want to be able to do what you're describing
jacky joined the channel
#
[tantek]
clicking on a photo would show the entire photo post, which presumably if you want to show it in context of other text, etc, then the content makes sense for that
#
[tantek]
either you're showing a photo-only view (use u-photo), or you're showing all the content (e-content)
#
[tantek]
so it's confusing why there was this push to show both in the reader at the same time as if they are separate things
#
aaronpk
because i want the reader to show the photos from all posts consistently rather than however they might appear inside HTML
#
aaronpk
what you're describing works if you only have: a view of photos with no content, and separately a view of content while ignoring the "photo" property
#
[tantek]
that's a MUCH harder problem. you don't get to pick how a publishers shows photos vs captions etc.
#
aaronpk
s/show the photos/show the photo posts/
#
[tantek]
this --> "show the photos from all posts consistently" <-- is totally a different kind of problem. "consistently" in what kind of context?
#
Loqi
this has -1 karma over the last year
#
aaronpk
consistently in the context of the reader
#
[tantek]
photo for preview, content for full display
#
aaronpk
like even just say you want to replicate the instagram UI, you can't do that right now without going through all the hoops i've had to do to remove the photo from the content
#
[tantek]
reformatting the content from an author is out of scope IMO, at least for any microformats / typical use-cases
#
aaronpk
"photo for preview, content for full display" is extremely limiting in building useful UIs
#
aaronpk
basically it forces me to treat every post as if it could always be a full blog post with arbitrary formatting and layout
#
[tantek]
yes, and I think that's an accurate view of what posts are!
#
aaronpk
not to mention then i have to special case the opposite situation where there is a photo property that doesn't appear in the content
#
[tantek]
yeah that's more weird
#
aaronpk
so you can't have it both ways
#
barnaby
is checking and removing duplicate photos from content really a lot of hassle? IIRC the URLs in e-content should be resolved by the mf parser and therefore guaranteed to be the same as the photo property, so it should be a case of checking for img or picture elements matching photo properties and removing them if you want to do something specific with the photos
#
[tantek]
it's not about "having it both ways". like I said, reformatting author published content is way out of scope for typical use-cases (and things people want to bother trying to code). it's also horribly fragile
#
aaronpk
barnaby: the simple case isn't that bad if you're willing to use an HTML parser, but then there are the edge cases
#
aaronpk
see this thread for an adventure https://github.com/aaronpk/XRay/issues/52
#
aaronpk
i never said anything about reformatting author content
#
Loqi
[aaronpk] #52 Remove images from posts containing a photo
#
[tantek]
I feel this is not going to get resolved again until we get an in-person IWC
#
barnaby
IMO every time you touch untrusted HTML content, there’s already so much sanitization steps which you need to jump through, that adding mf-specific behaviour like this isn’t a big deal
#
aaronpk
i feel this is never going to get resolved 😂
#
aaronpk
barnaby: read the thread, i'm stuck on some stuff with alt text and microformats right now :P
#
barnaby
and is necessary to handle the inevitable diversity of markup styles which are gonna show up in the real world
#
[tantek]
focusing on the "how do I remove u-photo from e-content" is completely missing the broader contexts of what publishers do what they do, and the simpler reader use-cases I listed above
#
aaronpk
[tantek]: what you described means i can't have my reader work the way it currently does
#
[tantek]
so we're not going to solve this with a singular issue thread like that. or rather any attempt to do so is going to disproportionately twist a design/solution to fit that and neglect the broader use-cases and ecosystem
#
aaronpk
so if that's how this gets resolved then i will be very unhappy but it will be possible to change it
#
[tantek]
and folks publishing photo posts will either care how their photo posts show up in readers which are doing simple but predictable things or they won't
#
aaronpk
and by "work the way it does" i mean the UI part, not the plumbing
#
[tantek]
the problem here AFAICT is jumping straight to narrow issue discussion without documenting the broader use-cases and how people publish what they do
#
aaronpk
i don't think that discussion has been narrow at all
#
aaronpk
it links out to like 5 other issues across a couple different repos too
#
barnaby
I definitely agree that mf2 plaintext parsing of alt text and image urls creates some problems
#
barnaby
it’s especially a pity that image alt text is separated from the u-photo property, resulting in consumers not being able to accurately recreate the original markup
#
barnaby
which could also be an accessibility issue
#
barnaby
but maybe this has already been discussed and I’m out of date?
#
aaronpk
no i think that is accurate
#
[tantek]
the discussion has been too plumbing-focused from what I've seen. e.g. we don't have screenshots for different use-cases for publishing & presenting in readers etc.
#
aaronpk
here's the remaining open issue that stemmed from that https://github.com/microformats/microformats2-parsing/issues/16
binyamin[m], rrix, vladimyr, LaBcasse[m] and ermahgerddavid[m joined the channel
#
[tantek]
barnaby, pretty sure the alt text in images has been handled for quite some time, in a way that makes it possible to recreate (i.e. they're kept together)
#
Loqi
[aaronpk] #16 consider not including img alt text as part of surrounding text properties
#
[tantek]
like I think we had to solve that for Micropub back in the day at W3C
#
aaronpk
we actually preemptively added it to micropub before it was resolved https://www.w3.org/TR/micropub/#uploading-a-photo-with-alt-text
jacky joined the channel
#
Loqi
[tantek] Resolution: issue 2 proposal accepted. No objections in above discussion, and positive opinions (👍) from a few implementors on the proposal. Proposal text incorporated into spec and reviewed. Proposal implementation in mf2py parser, test cases p...
jacky joined the channel
#
[tantek]
and yes we still need to fix #16
#
barnaby
ah, nice
#
barnaby
looks like that got added to php-mf2 in 1.4.7, and http://pin13.net/ is using 1.4.6
#
aaronpk
oh shoot really?
#
aaronpk
i thought it was up to date
#
barnaby
hmm actually it looks like there was only a 1.4.7 milestone, not an actual release yet
#
aaronpk
oh there is no 0.4.7 release yet right https://github.com/microformats/php-mf2/tags
#
barnaby
and the img alt parsing was part of the 1.4.7 milestone
#
barnaby
hmm last release was in 2018, maybe it’s time for a new one
#
barnaby
gotta be worth it if only for that parsing update
#
aaronpk
looks like we were planning on jumping to 0.5.0 https://github.com/microformats/php-mf2/milestone/4
#
barnaby
I’d go do it myself now but I don’t wanna step on people’s toes
#
aaronpk
i think everyone's toes are well out of the way
#
barnaby
lol maybe I’ll exercise my “original maintainer” priviledge then, move that one unfinished issue to 0.5.1, and make an 0.5 release
#
aaronpk
that would be much appreciated! tbh I don't even remember what exactly if anything we were waiting for
#
barnaby
okay I’ll go do that now
#
aaronpk
maybe gregor remembers the state of things?
#
barnaby
their post-libera irc nick is gRegor, right?
#
barnaby
!tell gRegor heya, any objections to me moving https://github.com/microformats/php-mf2/issues/184 to the 0.5.1 milestone and wrapping up the 0.5.0 release? it’s been four years since 0.4.6, and there’s easily enough features and fixes to justify a release IMO
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
[gRegorLove] #184 Fix vevent location backcompat
#
barnaby
maybe I’ll take the opportunity to set up some github actions on php-mf2 for testing and linting. I really like how the actions for my indieauth and micropub libraries turned out
#
aaronpk
oh yeah we lost those with travis didn't we
jacky joined the channel
#
[tantek]
barnaby++ a 0.5.0 release of php-mf2 would be amazing
#
Loqi
barnaby has 2 karma in this channel over the last year (8 in all channels)
#
barnaby
but github’s actions work really nicely, and are better integrated. installing dependencies, running tests and static analysis over multiple PHP versions is trivial and comes in a nice UI e.g. https://github.com/Taproot/indieauth/runs/4883719693?check_suite_focus=true
#
aaronpk
yeah i've set it up for some of my libraries to replace travis. it'd be great to get it back up on php-mf2 too
#
barnaby
jacky: did you make any progress on a rust mf2 parser?
#
jacky
barnaby: tremendous
#
jacky
test::generated::microformats_v2::h_event::concatenate
#
barnaby
cool! where can I find it?
#
jacky
it's not yet complete
#
jacky
I have two test cases of `h-event`, one of `h-review-aggregate` and two of `h-resume` to complete
#
barnaby
still using hard-coded property lists?
#
[schmarty]
hahaha php-mf2 stable will finally parse img alts? yaaaaaaassssss
#
Loqi
hahahaha
#
barnaby
ah, parser.rs and parse.rs contain separate parser implementations?
#
jacky
yeah - parse.rs is what I'm doing and parser.rs was the original code
#
barnaby
and parse.rs is the newer one?
#
jacky
I should delete that, heh
#
barnaby
got it. yeah it looks a lot better!
#
Loqi
it is probable
zack[m] joined the channel
#
barnaby
as an outsider poking through the code, my first instinct was to look in parser.rs, yeah
#
[schmarty]
omg i _just_ figured out that those are meant to be package namespaces and not websites lol
#
barnaby
or at least add a big // DEPRECATED, look in parse.rs ;)
#
jacky
just removed those files
#
jacky
haha [schmarty]
#
barnaby
[schmarty]: yes, soon! I will wait a day or two to give gRegor a chance to respond, then make a release
npd[m] joined the channel
#
[schmarty]
barnaby++
#
Loqi
barnaby has 3 karma in this channel over the last year (9 in all channels)
micahrl[m] joined the channel
#
barnaby
parse.rs looks good jacky++
#
Loqi
jacky has 21 karma in this channel over the last year (63 in all channels)
#
barnaby
I will read through it properly when I get a chance
#
jacky
def needs a lot more documentation
#
jacky
working on making some doctests to go with it too
#
barnaby
I can recommend embedding links to the relevant sections of https://microformats.org/wiki/microformats2-parsing by the code which handles each step
#
barnaby
makes it really easy to see which code is doing what, to double-check spec adherence, and to see if things are missing
#
jacky
ah good point
#
jacky
some parts of that doc aren't the easiest to link
#
jacky
tempted to litter it with <a name=""> lol
#
barnaby
hmm yeah or change the uls to ols?
#
barnaby
ideally synced up with the automatically numbered section headings
#
barnaby
e.g. so “• start with an empty JSON "items" array and "rels" & "rel-urls" hashes” would be 1.1.1
#
barnaby
but even just linking to the currently linkable section headings is useful
#
jacky
TIL re: doing that with <ol>
#
jacky
but yeah, I can do that!
#
jacky
tbh I came back to this b/c I wanted to start working on webmentions
#
jacky
and realized that this was really needed (imo) for it to be complete
#
barnaby
yeah, webmentions are boring without mf2 parsing capabilities
#
barnaby
my original motivation for php-mf2 was exactly the same
#
jacky
hopefully I can get this working to have something living at https://rust.microformats.io
#
jacky
which reminds me
#
jacky
how are those sites built? o.o
LaBcasse[m] joined the channel
#
jacky
ah nvm i see
#
barnaby
IIRC shane manages microformats.io
#
jacky
ah TIL
vladimyr joined the channel
#
aaronpk
not anymore, it's on the indieweb heroku
#
aaronpk
and we can definitely set up rust.* if you want
#
barnaby
ah okay
#
aaronpk
as long as it can run on heroku :)
#
barnaby
as always, assume my knowledge of the state of the community several years out of date xD
#
jacky
I can provide that once this is on https://crates.io tbh
ermahgerddavid[m joined the channel
#
aaronpk
When you're ready I can create a heroku project that you can deploy to if you'd like
diegov joined the channel
#
jacky
will do
#
jacky
ideally I wanna 'announce' that on my site :)
KartikPrabhu, reed, unrelentingtech, rommudoh[m] and Abhas[m] joined the channel
#
[tantek]
barnaby, surely you mean s/name=/id= 🙂
#
barnaby
that was jacky!
#
[tantek]
pretty sure we have folks in the #indieweb community younger than that. not naming names ofc 😉
cambridgeport90[, EvanBoehs[m], kandr3s[m] and juanchipro[m] joined the channel
#
barnaby
if the 1998 in the URL is accurate then the feature is younger than me, but my first HTML book definitely taught <a name="">
#
barnaby
struck me as a bit weird to use <a> for two completely different things even at the time
#
aaronpk
i thought the browser behavior of jumping down to the matching id= attribute based on the query string fragment was relatively newer than the actual id attribute
mambang[m] joined the channel
#
[tantek]
aaronpk it's been true since 1999 at least, we shipped it in IE4.5 Mac (if not IE4 Mac in 1998)
#
aaronpk
well i did start learning html well before that so...
#
capjamesg[d]
[tantek] I may or may not fall into that age group haha.
rommudoh[m], nekr0z, reed, Abhas[m], zack[m], samwilson, doubleloop[m], diegov, mambang[m], unrelentingtech, vladimyr, kinduff, binyamin[m], kandr3s[m], ermahgerddavid[m, juanchipro[m], cambridgeport90[, EvanBoehs[m], LaBcasse[m], npd[m], micahrl[m], Matt[m], jacky, angelo and tetov-irc joined the channel