#dev 2019-07-22

2019-07-22 UTC
KartikPrabhu, [dogeared], ben_thatmustbeme, [Michael_Beckwit, wagle, [xavierroy] and [tonz] joined the channel
#
Zegnat
!tell GWG,jgmac1106 I did experiment with having all posts (per mont) as fragment URLs: https://grapefruit.zegnat.net/
#
Loqi
Ok, I'll tell them that when I see them next
[Rose] joined the channel
#
jacky
the more I attempt to consider making video content
#
jacky
the more I realize I want something like youtube but either single-tenant or just planet-y
#
Zegnat
How far along is peertube with what you need, jacky?
#
jacky
It's not really what I'm looking. I think I might try this on my site first but being able to just post videos on my site and getting reactions for that might work for me.
[tonz] joined the channel
#
sknebel
jacky: what don't you like about peertube? Too standalone?
#
sknebel
I wondered about using a peertube instance and embedded videos on the main site, not sure if it supports that
#
GWG
Zegnat: I do know about your grapefruit site, but I forget how the various parsers and consumption tools dealt with it
#
Loqi
GWG: Zegnat left you a message 5 hours, 17 minutes ago: I did experiment with having all posts (per mont) as fragment URLs: https://grapefruit.zegnat.net/
#
Zegnat
Last time I was trying it didn’t work too well
#
Zegnat
XRay was able to parse posts, to some capacity, but it would only parse from the post element down meaning that data higher up (such as author) got discarded
#
GWG
My Parse This code seems to get it right.
#
jeremycherfas
What's the difference between PHP CGI and PHP FastCGI? And would it matter which I was using? I'm having trouble after upgrading Known and phpinfo says I have 7.0.32 while Dreamhost cpanel says I have 7.2 CGI
#
GWG
It doesn't however get a URL with a fragment right for individual posts
#
GWG
I should look into that
#
GWG
But the php-mf2 library has that issue with IDs
#
Loqi
[dshanske] #211 ParsefromID should include the element it is Parsing
#
Loqi
agreed.
#
Zegnat
jeremycherfas: FastCGI is the one I expect your server uses.
#
Zegnat
The difference is really just in how your webserver (Apache? Nginx?) talks to PHP to run code. Unless you are in charge of those configurations, you should not have to care
#
jeremycherfas
It was on plain CGI. I just switched to FCGI. But I am confused by the mismatch between phpinfo and cpanel.
[grantcodes] joined the channel
#
Zegnat
cPanel should be handling those pesky server settings for you, imo. So if those do not match, write to your hosting provider
#
Zegnat
If cPanel lets you decide between CGI and FastCGI ... pick whichever. It probably doesn’t matter on your scale at all.
#
jeremycherfas
All seems good now, thanks.
[Rose] and eli_oat joined the channel
#
@mattbd
Just stumbled across the Micropub (https://indieweb.org/Micropub) and Microsub (https://indieweb.org/Microsub), and really want to implement them. Will need to investigate to see how plausible that is for a static site...
(twitter.com/_/status/1153289156606537728)
[grantcodes] joined the channel
#
[grantcodes]
!tell [jgmac1106] Have you seen this? https://plaudit.pub/ I think it has webmentions hidden in there somewhere
#
Loqi
Ok, I'll tell them that when I see them next
[jgmac1106] joined the channel
#
[jgmac1106]
grantcodes I have not but I will reach out, ORCID to DOI would allow for some interesting stuff
#
Loqi
[jgmac1106]: Zegnat left you a message 7 hours, 41 minutes ago: I did experiment with having all posts (per mont) as fragment URLs: https://grapefruit.zegnat.net/
#
[grantcodes]
I found it from a node webmention library, so looks like it either uses webmentions or it will in the future in some way
#
[jgmac1106]
searching the code though I didn't find any use of endpoints
#
[jgmac1106]
reaching out to them now
#
[grantcodes]
It's amazing how little I understand what any of that stuff actually means. So much jargon and acronyms!
#
[jgmac1106]
ORCID=person unique identitider, a DOI is a docuements unique identifier
#
[jgmac1106]
My ORCID is 0000-0003-1035-389X
[snarfed], [schmarty] and [manton] joined the channel
#
[manton]
Some platforms (like WordPress or Micro.blog) have standalone "pages" that are treated differently than normal blog posts. Is there a convention for this in Micropub? This seems a little different than what people do for post types. WordPress's XML-RPC API has separate endpoints for "pages", which MarsEdit uses.
#
aaronpk
there isn't yet, but it's a good question
#
aaronpk
i do that in my micropub endpoint by having the client specify the slug with a leading /
#
[manton]
Ah, interesting.
#
aaronpk
the other question is whether h-entry is the most appropriate vocabulary or not
#
aaronpk
because typically those "pages" don't have published dates or authors
#
[manton]
The other side of this is being able to query a list of those kind of pages only.
#
[manton]
Right, I think not having a date (and not included in the usual blog post list) is what distinguishes these, at least the way WordPress and Micro.blog thinks about them.
#
aaronpk
i agree, same with my site
#
[grantcodes]
What do you actually want to do with micropub that is different from other posts?
#
aaronpk
specify that you're creating a page not a blog post
#
aaronpk
e.g. don't put this into the feed, don't give it a published date, give it this specific URL, etc
#
KartikPrabhu
how about if no "type" is specified it becomes a page?
Vincent joined the channel
#
[grantcodes]
Hmm I guess that would have to depend on the server implementation at the moment. Lots of options though. I would probably just set a category and have that handled specially
#
[schmarty]
it's interesting because micropub relies on microformats and a "page" is... kind of meaningless?
#
aaronpk
what properties does a "page" have other than content and name?
#
[schmarty]
lol grantcodes was just typing that
#
aaronpk
can a page have tags still?
#
[schmarty]
kartikprabhu++
#
Loqi
kartikprabhu has 2 karma in this channel over the last year (10 in all channels)
#
[manton]
One of my motivations for asking this: MarsEdit is a nice way on the Mac to edit pages, but it only shows the pages UI when you're using WordPress. I'd love to be able to suggest a way Daniel could enable this UI for more platforms than just WordPress.
#
aaronpk
it's interesting that there was never a microformat defined for this, whereas things like reviews or recipes got one early on
#
aaronpk
i suspect wordpress includes hEntry markup on pages too right?
#
aaronpk
it seems like pages have many properties in common with h-entry, tho definitely not the response properties such as like-of or bookmark-of
#
aaronpk
and the main distinguishing feature is that the post has no date
#
aaronpk
s/post/page
#
[manton]
Yes, I just created a test page on my test WP.com blog and it does include "hentry" for it.
#
aaronpk
so i wonder if the lack of the date could distinguish between an h-entry post vs page
#
[manton]
Back when I was using WordPress, I would sometimes create pages for RSVPs too, so they were separate from normal blog posts.
#
aaronpk
was that like a hack to make an unlisted post?
#
[manton]
Pretty much, yes.
#
[manton]
Maybe not relevant here, but just was thinking about mixing post kinds and these separate pages.
#
aaronpk
straw proposal: a micropub request with published=false indicates that it should create a "page", not put it in any feeds, and not assign a published date. this could also be queried somehow by asking the micropub endpoint to return undated posts.
#
aaronpk
only downside is it kind of sounds like creating an unpublished post (status=draft) as in the current brainstorming
#
KartikPrabhu
status=page ?
#
[manton]
For the date question, I personally don't use dates for pages, but technically WordPress and Micro.blog do give them a published date, even if it's hidden.
#
aaronpk
currently we have post-status=published and post-status=draft, would post-status=page make sense?
#
[grantcodes]
dont disagree but published=false definitely sounds like it should not be published
#
KartikPrabhu
I have seen pages that display "last edited on .... " to indicate how 'fresh' the content is
#
aaronpk
hm then the published date may not be a good distinguishing feature
#
aaronpk
what is a page?
#
Loqi
page is short for webpage on the IndieWeb, and can by default be marked up with h-entry (like body class=h-entry) for link-preview purposes unless the page serves a more specific purpose like a homepage or an event https://indieweb.org/page
#
[grantcodes]
Maybe visibility might make sense? It's not so far off an unlisted post?
#
[schmarty]
since the list of differences from an h-entry is so small, maybe a "page" is still an h-entry ?
#
[grantcodes]
Actually that makes more sense to me, because unlisted pages definitely aren't a thing (private maybe...), but pages can definitely have a `post-status` of published, or draft, so that is a definite conflict
#
[schmarty]
the differences sound more CMS-based
#
[schmarty]
like "a page doesn't appear in feeds" (except on some people's sites they do)
#
[schmarty]
i'm not hearing any differences in micropub terms
#
aaronpk
[schmarty]: the difference needed in micropub is the ability to tell the server to create a page vs put the post in the main feed of posts
#
aaronpk
and to query the server for these posts separately from others
#
aaronpk
i don't think mixing it with visibility is right
#
aaronpk
a page may be linked in some navigation, which would go against the idea of an "unlisted" post
#
[manton]
Yeah, pages are often included in navigation, but not always. On my blog I have several pages that I hide from the navigation but which I still want to link to manually elsewhere.
#
aaronpk
for sure, i just don't think it's a good idea to mix permission stuff like visibility with this idea
chrisaldrich joined the channel
#
[manton]
Agreed. Someone will inevitably want to a private or unlisted page.
#
aaronpk
but yeah I agree that other than the lack of response properties (like-of, bookmark-of, etc), pages are probably best represented as h-entry
#
sknebel
Our wiki pages are too
#
aaronpk
oh right haha
eli_oat joined the channel
#
aaronpk
[manton] what do you think about post-status=page?
#
[manton]
I'd be fine with that because Micro.blog pages can't be drafts. But that would be the downside, if someone wanted a draft page.
#
aaronpk
heh post-status=draft-page?
#
[manton]
Using post-status seems better than visibility or published to me, though.
#
[manton]
Heh, yeah, I guess that's a solution if someone really needs it.
#
sknebel
Post status feels weird
#
[grantcodes]
If I used micropub for pages I would definitely want drafts for pages. Also backed up by the majority of wordpress sites I work on having draft pages
gRegorLove and [snarfed] joined the channel
#
[snarfed]
late to the conversation, but i use pages a lot on wordpress, as well as posts
#
[snarfed]
i'm not sure we need or want anything mf2-visible for page vs post
#
[snarfed]
wordpress at least has gradually converged the two over time, to the point where the difference there is arguably just an internal plumbing detail that mostly doesn't matter
#
aaronpk
visible mf2 isn't the question here
#
[snarfed]
eg my pages have normal h-entry mf2, author, published and updated dates, etc
#
[snarfed]
sorry, micropub
#
[snarfed]
only real difference on my site is whether the slug has a date in it
#
aaronpk
micropub explicitly needs something to handle this case
#
aaronpk
that "something" could be an instruction to set the slug to something without a date, and to not put the post in feeds, but something is needed
#
[snarfed]
ok! visibility, post status, url/slug, etc all seem good. tags and other ways to determine which feed(s) to put it in also seem good. clear user-visible features. all probably better than "page" vs "post"
#
[snarfed]
(which seems like a server-specific feature in m.b and WP at least)
#
[snarfed]
but you all are closer to it than me, i should shut up
#
aaronpk
in any case, there's been enough discussion here that it's worth capturing all the current examples on https://indieweb.org/page
#
[grantcodes]
🤔 random one, what if that was just how you handled micropub entries that are marked as unlisted but with a defined slug?
#
KartikPrabhu
what's "marked as unlisted" in micropub?
#
aaronpk
that does mean there is no distinction between listed and unlisted pages, so we'd want to survey whether that is ok
#
aaronpk
but afaik pages typically don't get put into lists automatically, only explicitly added to navigation
#
aaronpk
and it does slightly change the meaning of visibility to no longer imply permissions
#
GWG
Catching up on conversation
#
GWG
I agree with all people have said, and like unlisted
#
GWG
Although WordPress has a post status for that as well
#
[grantcodes]
visibility and post-status could probably be one property tbh. I can't see any overlap myself
[eddie] joined the channel
#
[eddie]
Also something to think about: If the difference between a page and a "post" is all about Micropub and not about mf2, should it instead be mp-type: page or post? or something similar?
#
[eddie]
It sounds more like an mp- server command then an attribute about the post itself
KartikPrabhu joined the channel
#
GWG
Visibility refers to how it is accessed. Status refers to whether it can be accessed
#
aaronpk
[eddie]: ooh interesting thought
#
GWG
Now, admittedly WordPress has two different settings, but stores in the same single variable
#
aaronpk
and yes GWG is right about visibility vs status
#
[grantcodes]
I think this is probably the point where implied post type discovery falls down. Wouldn't be an issue if you could explicitly set a post type, which sounds like what eddie suggested
#
aaronpk
Post type discovery only comes into the picture if a consumer needs to know the difference between a post and a page
#
GWG
When might they?
#
[eddie]
Exactly. I think PTD is fine. I don't think a consumer would necessarily need to determine the difference between a post and a page?
#
aaronpk
If a page doesn't show a published date then the consumer already isn't going to show a published date when showing the comment right? That's the only thing I can think of that'd be different really
#
[grantcodes]
Might just be my view of it, but I see a micropub server as a consumer of mf2, and this problem is the server attempting to discover if is a post or page
#
aaronpk
If you got a webmention from a page, would you need to show anything different?
#
aaronpk
Yeah that is kind of true [grantcodes]
#
aaronpk
the other way to think about it is there isn't a distinction between posts and pages
#
GWG
There isn't
#
aaronpk
there's just an instruction to the micropub server to do stuff with dates and URLs and putting the post in feeds
#
GWG
The date thing really doesn't apply
#
j4y_funabashi[m]
Wouldn't PTD come into play when a client wants to query for all "pages"?
#
GWG
Just its feed behavior
#
aaronpk
j4y_funabashi[m]: no, PTD is about determining the type of one thing
#
[grantcodes]
But agreed, there is no need to know if it is a post or page for most other consuming cases.
#
[grantcodes]
But there is an experimental post-type query for micropub that j4y_funabashi might be thinking of
#
aaronpk
Other consuming question is what does a reader do if it encounters one of these things in a feed? Tho by definition they don't exist in feeds so that probably isn't important
#
[eddie]
if someone added a "page" into a feed, I would expect it to look like an article
#
[eddie]
or a note
#
aaronpk
Yeah I can't think of anything else it should look like
#
j4y_funabashi[m]
If I q=source type=page my server needs to determine which posts are pages
#
aaronpk
nobody said type=page yet
#
aaronpk
yes, micropub query needs to be able to query for these
#
aaronpk
but that isn't necessarily a "type"
#
j4y_funabashi[m]
So an mp* would help with the publishing side but not the querying side
#
[eddie]
I think you could query for pages by matching a query to the mp-command, so that doesn't seem like an issue
#
[eddie]
I wonder if from a simplification side of things we're literally just telling the server if it's a page or not
#
[eddie]
If you do something like mp-page=true or false, then you could also add q=source&page=false
#
[eddie]
oops you would want q=source&page=true
#
aaronpk
We need some term that is shorthand for the features that these posts have
#
[eddie]
so the mp- command and the Micropub query could match and it seems pretty easily understandable
#
[eddie]
That makes sense. So I think right now we're at: it's unlisted and has a static url based on a slug?
#
[eddie]
I don't think dates are involved, as I include dt-published and dt-updated on pages
#
[eddie]
and I think things like a wiki page you might want dates
#
aaronpk
Yeah sounds like implementations are mixed on the date issue
#
[grantcodes]
I will say now that I've added post type queries to my reader and server I would totally set up pages as a post type just for the UI benefits. I've sort of done it for my journal posts and it's awesome
[tantek] and [fluffy] joined the channel
#
[tantek]
wow what a topsyturvy journey about pages
#
[tantek]
looks like most things have settled
#
[tantek]
pages are absolutely ok to markup with h-entry. *including* dt-published and dt-updated, as those are directly related to user actions (as properties should be if possible!0
#
[tantek]
dt-published - the point in time that the user first publishes the "page" on their site such that is visible at its u-url
#
[tantek]
dt-updated - any future time that the user "saves" the page
#
[tantek]
the question of "should it show up in feeds or not" has nothing to do with post vs page
#
[tantek]
there are /unlisted posts
#
[tantek]
and there are feeds that have pages when they are created or updated
#
[tantek]
the date issue is a matter of user presentation preference. that is, whether or not a page "has" a date is *purely* presentational. from a *literal* (physics) perspective, a page absolutely has both a date when it is first published, and if/when it is changed, a date of when that (latest) change occured
#
[tantek]
whether or not an author/publisher chooses to display that/those date(s) is purely presentational. from a content, CMS, backend, metadata, protocol perspective, pages have dates
#
[tantek]
(plenty of prior art, before blogging / journaling was even a thing, of web pages that had e.g. "Last Modified:" in their footer)
#
[tantek]
[eddie]'s latest point is the closest to making a distinction, at least from a "common patterns" perspective: permalinks for "pages" tend to not include date information, whereas permalinks for "posts" tend to have some date (year at least) information
#
[tantek]
though I've seen exceptions to both of those. there are blogs that just use domain-name/long-slug-with-all-the-seo-keywords for post permalinks
#
[tantek]
and there are (much fewer) sites that use domain-name/year/name-or-slug for pages (e.g. w3.org)
laureng|Microspo, [JWvdPol], [jgmac1106], Magnus[m]1, [schmarty] and leg joined the channel
#
[tantek]
Zegnat, do you markup your resume with h-resume? http://microformats.org/wiki/h-resume
#
Zegnat
No, it is only available as PDF currently. As that was a requirement during my last job hunt.
#
KartikPrabhu
I have an HTML resume and then I print it into a PDF when needed :P
leg1 and [manton] joined the channel
#
Zegnat
I had that for a while too, because HTML > .doc, but even then I never added the mf2 as I never published the html.
#
[jgmac1106]
My resume is h-resume. I had to use to different stylesheets bc I couldn't figure out how to write a print media query where <details> defaults open
#
[tantek]
Requirement for format B does not preclude format A
#
[tantek]
just like requirement to post on social media does not preclude posting on your own site
#
KartikPrabhu
[jgmac1106]: oo that is an interesting print CSS case!
#
[jgmac1106]
Kinda funny when ever we talk tracking annual progress the IT staff jealous about how well ny vita parses
#
Zegnat
Hmm, yeah, details sounds tricky.
#
[jgmac1106]
Yeah someone linked to an existing issue in the CSS repo and I wasn't exactly sure if you can't do it or if I didn't know how
#
Zegnat
AFAIK you can't do it. But I'd have to double check to give a definitive answer.
#
[jgmac1106]
Off topic from original h-resume thread but print query and detail auto open should be a thing, if it is, please point it to me so I can write a print query rather than two stylesheets
#
[jgmac1106]
What I thought. And looks like issue already open in GitHub
#
[jgmac1106]
My intern is working on a vita builder in Kirby that will produce an h-resume
#
KartikPrabhu
My first guess was to use ::after pseduo-element but you can't access the contents of the <details> from CSS
#
[jgmac1106]
Zegnat harkens back to the APA h-cite stuff we star5ed with
#
[jgmac1106]
@kartikPrabhu yeah and that gets way past my CSS knowledge in general
[KevinMarks] joined the channel
#
[KevinMarks]
You can set details to open in css though not sure if you can dynamically
#
[fluffy]
Of course there’s an h-resume
#
[fluffy]
I guess I should add that to my resume
#
[fluffy]
although I have to ask myself, why? the recruiter’s just going to print it out anyway.
#
Zegnat
[KevinMarks] you can? How?
#
KartikPrabhu
yeah, this I want to know
#
[fluffy]
I haven’t seen how to do it in CSS, just in HTML
#
[fluffy]
you can do e.g. <details open>
#
[fluffy]
or maybe it’s opened? I forget
#
[fluffy]
I was using <details> on my resume until people complained that they didn’t see the open widget and thought I was just being brief.
#
[KevinMarks]
Details open
#
KartikPrabhu
you can toggle that with JS, but how in CSS?
#
[fluffy]
I’ve had a hard time finding any decent docs on how to style <details> in the first place though
#
[fluffy]
lots of overly-verbose examples though
#
Zegnat
Sadly I don't think there is print-only JS, else this would be trivial, haha
#
Zegnat
Back to the days of displaying a print button on your site that triggers some JS first?
#
Zegnat
That attribute is HTML, though, [KevinMarks]. That's the issue here: want CSS to state whether it is open or not
#
[fluffy]
yeah so like a @media print or whatever can open everything
#
KartikPrabhu
right. CSS can find if details is open using details[open] selector. but you can't use that to set things to open by defualt
#
[fluffy]
the details spec seems a bit half-baked at the moment 😕
#
[fluffy]
like there doesn’t seem to be any standard way to style the marker as far as I’m finding, there’s just like a `::-webkit-details-marker` extension pseudoelement
#
Zegnat
There is no standard for that one yet either, nl
#
Zegnat
This is probably the issue jgmac1106 was referring to, touched on several things that are hard to style about details: https://github.com/w3c/csswg-drafts/issues/2084
#
Loqi
[patrickhlauke] #2084 Ability to style <details>/<summary> to be open (e.g. for print styles)
#
[fluffy]
https://github.com/w3c/csswg-drafts/issues/2084#issuecomment-351177835 was exactly where my brain was heading before I read that thread, so thanks
#
Loqi
[tabatkins] I'm down with that approach. The UA stylesheet is then extremely trivial: ``` details::details-content { display: none; } details[open]::details-content { display: contents; } ``` And overriding it is super easy. :+1: from me.
#
[KevinMarks]
Could you wrap the non summary part of details in an element and make that display block for print , or would the details override that?
[snarfed] joined the channel
#
KartikPrabhu
[KevinMarks]: no, the <details> element actually controls the display!
#
KartikPrabhu
I tried targetting the children of <details> and setting the display property but no luck
#
sknebel
"display:contents" on the <details>, and some extra styling on the <summary> to fix the open/close indicator
#
KartikPrabhu
sknebel++ that works!!
#
Loqi
sknebel has 43 karma in this channel over the last year (111 in all channels)
#
Zegnat
Going to have to test that tomorrow.
#
KartikPrabhu
you can get rid of the indicator on summary with display: inline or inline-block too
#
sknebel
+ if you want onPrint JS to handle browser that don't do display:contents yet
#
KartikPrabhu
and now it depends on the display: contents support :P
#
sknebel
that was what we found last time this came up :D
#
@stackpile
Weekly Platform News: CSS ::marker pseudo-element, pre-rendering web components, adding Webmention to your site https://sb.gl/2YW1pee
(twitter.com/_/status/1153407891493273602)
[grantcodes] and [chrisaldrich] joined the channel
#
[chrisaldrich]
I'll have to follow the details conversation more closely when I'm back in the office as it may have the solution, but I remember having an issue with it working in conjunction with fragments a few weeks back.
KartikPrabhu, [fluffy], AndrewB7794[m] and [KevinMarks] joined the channel