[eddie]For a while I had hard-coded my RSS feed to only display certain post types in my micro.blog rss file. But, then I had an idea. I made my rss file show any post that has it’s syndication property set to micro.blog
[eddie]This didn’t change much until yesterday because I hadn’t changed what types of stuff I posted until I decided I wanted to post *select* watch posts
[eddie]Because there is no data driven reason to show certain watch posts, but I did think “I thought this episode was particularly good and want to recommend it”
[eddie]It does not. Currently the home page is still running based on existing rules. Although that is run by a featured setting so I could add that (if syndicate to m.blog mark as featured). However I’m still deciding what I want to do with my homepage. I think I’m gonna take it in a different direction
[eddie]I’m thinking about having a “latest podcast” widget”, maybe shortening how many articles I display and then link to an archive. So I’m almost thinking of building different widgets for type of content rather than a stream
Loqi[Aaron Parecki] Encountered two blockers working on this:
1) In a simple example of an img tag inside an e-content tag, the parsers are using the img tag as an implied photo property. This seems wrong to me. Example This means XRay sees a post like this as a photo...
Zegnataaronpk: “since XRay can't be sure exactly what method was used to generate the plaintext value from the original HTML” - DOM’s textContent should be used, IIRC, else the parser is broken.
aaronpkthat function mostly uses the DOM textContent, but it has special handling for whitespace and certain html tags that weren't being handled right by the plain textContent function
ZegnatEither way, I would still say that when mf2 spec says “textContent” it means textContent per DOM spec. So XRay should be able to be certain of what the parser returns.
aaronpkany thoughts on my best path forward with this? I think I have to implement HTML->text conversion in XRay even though the mf2 parser is already doing it
[eddie]I feel like I somehow must have gone crazy. When I wrote the Jf2Post object in Swift, I added a “description” field. But it doesn’t seem that XRay or Microsub return description, they return content…. 😕 is description an older jf2 attribute name?
Zegnatcweiske DOMDocument says it exposes the DOM API. I would expect that to be the DOM spec. textContent for HTML does not extend on DOM textContent I believe.
aaronpkquestion: in a case like <a href=""><img src="..." class="u-photo"></a> where XRay removes the img tag, shouldn't it also remove the <a> tag since the tag is now empty?
snarfedi compared the activitypub POST with another one that succeeded (against a different instance), and the only meaningful differences are a tag and a location
snarfedhi [cleverdevil]! hey i expect you're press savvy...we might try to drum up a press cycle for https://indieweb.org/Webm1m . probably throwing a party in 2w. feel free to ping j12t if you can help!
ZegnatActually, I think Let’s Encrypt monitors for you, which is beautiful. If my renewal (which is set to do it far in advance of the cert lapsing) fails, LE is still going to email me about my certificate running out :)
LoqiPost Type Discovery specifies an algorithm for determining the type of a post by what properties it has and potentially what value(s) they have, which helps avoid the need for explicit post types that are being abandoned by modern post creation UIs https://indieweb.org/PTD
aaronpkoh yeah, I should clarify: the problem when there is no alt attribute is that the URL ends up in the content.text value. it is able to dedupe the name/content fine.
j12t_, j12t__, j12t___ and j12t joined the channel
ZegnatI do like the text browser output for plain text, and wouldn’t mind seeing that standardised somewhere, aaronpk. Even within mf2. But it is definitely hard and you will always end up with opinionated choices (and thus people who have to then go and build their own alternatives).
[kevinmarks]Why don't you dedupe the image after parsing rather than change what you are parsing? Or are you getting 2 different image urls due to thumbnails?
ZegnatTo be fair, those posts will not result in duplicated photos aaronpk. As the markup author specifically says the content does not contain images by using p-
ZegnatNo, I mean, the markup says: “this post has photo X and content Y, but the author does not want to show the photo within content because they specified to take content as plain text only”
ZegnatI guess the answer there is that if your deduping depends on alt values (which apparently it does), you are not allowed to remove alt values from the content :P
ZegnatI think in general markup around photos is just going to be hugely varied. Which isn’t going to help. People simply haven’t decided on how to mark up photo posts yet.
ZegnatI totally see the reason for the URLs. That’s actually one of the small parts I would expect from a plain text browsing experience. But I also see why they are annoying when you get only the mf2 parser output and not the original DOM, because you have no idea of the URL source.
[keithjgrant]is anybody using as static site generator to do recipes? I'd love to know if there's a workaround to avoid putting everything in the front matter... I can't figure out how else to make it parseable by my template (to markup p-ingredients, especially)
aaronpkZegnat: the point is XRay will already show you the url even in a plaintext experience because it's in the u-photo property. I don't want it duplicated in the content still
[kevinmarks] I found that google will enable SSL for appengine apps for free now
bear for anyone who depends on LetsEncrypt and uses SNI validation for their certs -- https://community.letsencrypt.org/t/2018-01-11-update-regarding-acme-tls-sni-and-shared-hosting-infrastructure/50188
snarfed [kevinmarks]: hu. yeah app engine's LE integration had a few bugs at first but seems to work well now. maybe you haven't added www explicitly on https://console.cloud.google.com/appengine/settings/domains ?
Loqi red beating heart https://www.svgshare.com/i/4uZ.svg
tantek hey @all tomorrow is the W3C Webmention spec's one year birthday! https://www.w3.org/TR/2017/REC-webmention-20170112/ (aaronpk)
tantek.com edited /press-kit (-103) "active leaders, at least participated, preferably have led/organized HWC/IWC in the past year or two" (view diff)
@martijnvdven Congratulations Webmention, on turning 1 year old as a W3C Rec 🎉
But age isn’t the only milestone worth celebrating… https://twitter.com/i/web/status/951748459698192384 (twitter.com/_/status/951748459698192384)
jeremycherfas.net edited /User:Vanderven.se_martijn/hwc-template (-37) "Tidy up intro" (view diff)
[eddie] It does not. Currently the home page is still running based on existing rules. Although that is run by a featured setting so I could add that (if syndicate to m.blog mark as featured). However I’m still deciding what I want to do with my homepage. I think I’m gonna take it in a different direction
aaronpk "Implemented @glennjones "innerText" parsing for better handling of block level elements and whitespace." https://github.com/indieweb/php-mf2/commit/cc3037200d33455bac522c423fb14d81478c1313
Zegnat Does tantek have explicit names on those posts, or are those implied names and will they disappear if we resolve https://github.com/microformats/microformats2-parsing/issues/6 ?
gRegorLove uh oh, glennjones' tls cert expired https://glennjones.net/tools/microformats/
KartikPrabhu what is PTD?
jonnybarnes so trying to add better supports for likes
[keithjgrant] is anybody using as static site generator to do recipes? I'd love to know if there's a workaround to avoid putting everything in the front matter... I can't figure out how else to make it parseable by my template (to markup p-ingredients, especially)