#dev 2017-11-12

2017-11-12 UTC
snarfed, tantek and [keithjgrant] joined the channel
#
aaronpk
I don't think I have energy left to continue being productive today. I did get a lot done this morning at least.
#
GWG
Want to share, aaronpk?
#
GWG
I opened an issue to refactor my living room, committed some changes, but I need to commit more before I declare a stable branch.
#
[miklb]
I did see a new release for x-ray in indenews
#
[miklb]
indienews feed
davidmead and snarfed joined the channel
#
Loqi
[aaronpk] emoji-detector-php: This library will find all emoji in an input string and return information about each emoji character.
davidmead joined the channel
#
snarfed
i guess their use cases are a bit different
#
GWG
I've never used either of them. I considered using them in Semantic Linkbacks to improve reaction support from Bridgy, but never did.
tantek joined the channel
#
Loqi
[dshanske] #60 Reacji
#
aaronpk
That reminds me, I have a ZWJ bug on that I need to look into. Might just need a Regex update.
#
GWG
snarfed: I wish you good luck. I am glad to see contributions to Semantic Linkbacks. As you recall, I did a round in the summer and will likely go through again.
#
GWG
snarfed: I am close to making Post Kinds MF2 JSon compatible so that it will work better with the Micropub plugin
snarfed joined the channel
#
Loqi
gwg has 16 karma in this channel (270 overall)
#
GWG
snarfed: All the Checkin data I filed while abroad helped me work on the testing. It's also why I added the Swarm icon to Syndication Links.
#
GWG
I have a bunch more tweaks, because this changes the meta key storage. But I've been so busy with life of late.
tantek and [miklb] joined the channel
#
[miklb]
I still get confused by the semantic linkbacks plugin name and confuse it with syndication links.
#
GWG
[miklb]: Well, we always talk about merging it into Webmentions
#
[miklb]
I recall the discussions and arguments for/against merging.
#
GWG
[miklb]: I think pfefferle left it as if it ever got to be less hacky, he'd withdraw his concerns. I think we keep evolving it and it may get there.
#
[miklb]
sure. understood. I’ll still confuse the plugins until then ?
renem joined the channel
#
tantek
agreed with [miklb] - the name is confusing
#
tantek
speaking of confusing, I forgot how I post video
#
GWG
tantek: I think the name is staying as long as the plugin does. The better move is a merge.
#
tantek
I think I need to improve /Falcon documentation just for user 01. ?
#
GWG
Although, I'm pleased snarfed is taking an interest. That moves things along
#
GWG
tantek: Documentation for yourself is good.
snarfed, tantek, [eddie], tbbrown, gRegorLove and [asuh] joined the channel
jeremycherfas joined the channel
#
jkphl.is
uploaded /File:iwcber-2017-group.gif "IndieWebCamp Berlin 2017 group photo (animated) https://indieweb.org/File:iwcber-2017-group.gif"
[dgold] joined the channel
#
seblog.nl
created /Cineville (+916) "Created page with "{{stub}} '''<dfn>[https://cineville.nl Cineville]</dfn>''' is a Dutch network of art house cinema's, a card with a monthly subscription to see unlimited movies in those cinema's...""
(view diff)
#
Zegnat
Does anyone here have any experience setting up Caddy?
#
@keithjgrant
@rachsmithtweets I added webmention support to my blog and I love it. And it's a W3C Recomendation spec!
(twitter.com/_/status/929681954370400256)
loicm, raretrack and [keithjgrant] joined the channel
#
[keithjgrant]
If I reply to a reply, should my in-reply-to lost both urls?
#
Zegnat
The in-reply-to goes for the specific post you are replying to. You could be replying to both posts, but if you are writing your reply purely because of 1 post then that 1 post should get the in-reply-to. Whether that 1 post happens to be a reply itself doesn’t affect anything.
raretrack joined the channel
#
sebsel
[keithjgrant] See also Salmentions for notifications about replies on replies
#
sebsel
what is salmention?
#
Loqi
Salmentions are a protocol extension to Webmention to propagate comments and other interactions upstream by sending a webmention from a response to the original post when the response itself receives a response (comment, like, etc.) https://indieweb.org/salmention
#
sebsel
Not widely supported / implemented yet though.
#
aaronpk
attempts to install php 7.1 on his laptop to see how much stuff breaks
#
aaronpk
I figure it's about time to start thinking about moving off of 5.6, since there's only a year left of security updates
#
aaronpk
the real question is whether I start making my open source php apps work with just 7.x or try to continue supporting 5.6
#
Zegnat
For everything that isn’t a library (like mf2 parser) I would say going to 7.x is fine.
#
Zegnat
Apps that are supposed to be hosted whole scale should be fine on 7.x, people interested in hosting those specific apps should look at hosting that works for them
#
aaronpk
that's kind of what i'm thinking
#
aaronpk
for libraries i'd want to support as far back as is reasonable
#
loqi.me
edited /Jekyll (+127) "aaronpk added "[https://www.aaron-gustafson.com/notebook/enabling-webmentions-in-jekyll/ Enabling Webmentions in Jekyll] by Aaron Gustafson" to "See Also""
(view diff)
#
aaronparecki.com
edited /Jekyll (+46) "/* See Also */"
(view diff)
snarfed joined the channel
#
aaronpk
so far so good. my home page works fine in 7.1
John___ joined the channel
#
aaronpk
oh boy
#
aaronpk
something isn't happy with the imagemagick library
#
aaronpk
uninstalling and reinstalling imagemagick from source fixed it. *whew*
[kevinmarks] joined the channel
#
Zegnat
Is Imagemagick ever actually happy?
#
[kevinmarks]
dgold I'd have to check, but I think both provide a script to fetch webmentions and display them
#
sebsel
dgold how do you display them at the moment? you already fetch them somehow, right?
#
dgold
yes, using a javascript that voxpelli helped me with
#
sebsel
Ah, so you don't even store them in your generated html... you let your visitor fetch them from webmention.herokuapp.com
#
sebsel
I don't know if webmention.io has a display like that.
#
dgold
this is possibly not the most appropriate way to do this, but it was the only way I could get it to work
#
sebsel
Tantek has a bigger hack :P
#
Zegnat
I was just going to say tantek has a similar set-up. These make it very hard to switch services.
#
sebsel
Well, in a way, both are not really owning their received webmentions, because they rely on third party services to store them.
#
sebsel
not saying that it's totally wrong. But it's hard to move your data that way
#
dgold
very much so, which is why it was the next stop on my indieweb journey
#
sebsel
remember that you can always re-send old webmentions to the new service.
#
aaronpk
webmention.io doesn't have a display like that partly because I'd rather people didn't rely on it to store their webmentions :-P
#
dgold
but I'm running into a brick wall, there seems to be a dearth of implementations available to peruse
#
aaronpk
it has an api so you can download them, and I recently added a web hook so you can get them sent to you in real time
#
sebsel
sees room for a webmention-storage service with easy export
#
Zegnat
You could also make a webmention-replay service. E.g. hook into webmention.io API and resend all the webmentions to your new storage
#
snarfed
sebsel: "self-hosted all the way down" is explicitly not an indieweb design goal. all of our sites rely on third party services at some point, whether it's hosting, datacenters, backbone network providers, etc
tantek joined the channel
#
snarfed
we each strive for *some* degree of portability, usually more than the silos, but we obviously don't go all the way and write our own web servers, OSes, etc all from scratch. where we each draw the line is a personal choice, right?
#
snarfed
i think we've captured that well in e.g. https://indieweb.org/web_hosting#Criteria
#
sebsel
snarfed I do realise that!
#
snarfed
sebsel++
#
Loqi
sebsel has 16 karma in this channel (51 overall)
#
Zegnat
I think it is less about hosting everything yourself and more about portability. If we are seeing a portability issue amongst IndieWeb services (in this case webmention endpoint providers) that is something to talk about :)
raretrack joined the channel
#
snarfed
Zegnat++
#
Loqi
zegnat has 35 karma in this channel (143 overall)
#
snarfed
portability indeed. arguably the only truly required thing to be indieweb is a domain. how far to go beyond that is personal choice. (ie even tumblr, blogger, and wordpress.com are viable indieweb hosts.)
#
tantek
Zegnat I agree with the importance of but I'm not sure about the focus on portability. Day-to-day it's not the biggest factor IMO.
#
tantek
what is portability
#
Loqi
It looks like we don't have a page for "portability" yet. Would you like to create it?
#
loqi.me
created /portability (+29) "prompted by tantek and dfn added by tantek"
(view diff)
#
tantek
Zegnat, perhaps you can find some indieweb examples / experiences of portability and add them there? e.g. people switching from static/Jekyll to/from WordPress to/from Known?
#
Zegnat
I am not sure it is a focus per se. But as we like to talk about whether social media silos support proper exports, we shouldn’t forget taking a look at our own silos like the webmention endpoint providers. I was talking specifically about portability in that sense, not really about switching CMS.
#
Zegnat
But switching CMS is definitely also interesting. I’ll give a think about who has made moves like that
#
sebsel
tantek It's not the biggest factor, day to day, but we just had dgold struggling with the portability of his received webmentions.
#
dgold
well, that's the primary reason I use an SSG - all posts are stored as flat files, json/YAML frontmatter
#
dgold
there's no data portability issue, if a new shiny SSG comes along, converting the frontmatter is a quick script away
#
tantek
dgold, do you store the responses as well? or do you depend on a service to show those in realtime?
#
dgold
tantek: I rely on a service, in my case voxpelli's
#
aaronpk
this is how the conversation started :)
#
tantek
yes, flat files help A LOT with portability
#
dgold
portability + database avoidance = win*
#
dgold
*webmentions notwithstanding
#
tantek
interesting. assuming you can iterate all the webmentions ever received by a service on your behalf, you can 1) switch your webmention endpoint, 2) iterate and literally resend all webmentions
#
tantek
^^^ that's how you do cheap & quick webmention service portability
#
tantek
the protocol was designed to make that possible
#
aaronpk
yep and that also doesn't require that either webmention service support importing, only relies on being able to list all received webmentions, which presumably it has to be able to do to be a useful webmention service
#
tantek
precisely
#
tantek
your webmention endpoint is the importer
#
Zegnat
Yep, which is why I mentioned doing a replay service ;)
#
sebsel
the only problem is that some pages will 404 (or similar) after a while, in which case the old endpoint might have a cached version, but the new one doesn't.
#
sebsel
But one can argue that in case of a non-200, you should let go of your cache.
#
tantek
sebsel - good reason to send every webmention received to the Internet Archive
#
Zegnat
Theoretically, if the page has gone unavailable, the original sender should have sent an update webmention and you would have taken it out of your cache in the first place, right?
#
tantek
not for a 404 no. we've already had that debate. 404s from received webmentions are better treated as transient errors
#
tantek
no. only for 410
#
tantek
that's in the FAQ I'm pretty sure
#
Zegnat
Well, that’s a discussion for people who actually want to display webmentions. I do not, so I will leave that up to other to decide, haha.
#
Loqi
ahaha
#
sebsel
yeah, that's one we had earlier today too: mention on a page != webmention != displaying received webmentions
#
sebsel
I believe the thing that dgold wants to save here is the display of received webmentions?
#
tantek
Zegnat, it's only a reasonable discussion if there is new information beyond what is already in the spec / existing discussion. Otherwise rather than "leave it", it's better to cite the current answer as a starting point: https://indieweb.org/deleted#404_Discussion
#
www.svenknebel.de
edited /Jekyll (+99) "/* Related Tools */ details about jekyll-webmention.io"
(view diff)
#
Zegnat
I meant that the entire discussion doesn’t apply to me in the first place tantek, so leaving it to others feels 100% like the right thing for me to do
#
tantek
Zegnat, no, to be helpful, it's always nice to cite existing answers on the wiki, even if it's for stuff you don't do
#
tantek
to help people not waste time re-hashing
[miklb] joined the channel
#
tantek
hmm I wonder how hard it would be to write a mini-pass through webmention handler
#
tantek
e.g. receive the webmention, do something very simple (e.g. send a copy to Internet Archive), then pass through to a webmention handler service
#
tantek
though maybe only verified webmentions should be archived? and then you're doing much of the work of the service anyway
#
aaronpk
that's essentially what webmention.io is
#
aaronpk
it would be trivial for me to add an option in webmention.io to have it ping web.archive.org
#
tantek
why not make it a default?
#
tantek
too much traffic?
#
tantek
presumably webmention.io is handling only public webmentions now anyway right?
#
aaronpk
it can handle private webmentions
#
aaronpk
obviously web archive can't fetch those but I would just not ping for those
#
aaronpk
I think I would also exclude anything from brid-gy.appspot.com
#
tantek
to avoid hitting bridgy too much?
#
tantek
colinwalker++ for https://colinwalker.blog/2017/11/12/learning-from-the-past/ <-- definitely worth reading for all indieweb devs
#
Loqi
colinwalker has 1 karma in this channel (4 overall)
#
Loqi
[Colin Walker] Learning from the past
#
aaronpk
well and those permalinks don't look very good, and I don't think anyone would think to look up bridgy URLs to find old tweets
#
tantek
better reason: twitter is mostly dumpster fire now anyway?
#
aaronpk
oh and there's not a lot of value of archiving the "like" or "repost" pages either
#
Zegnat
I think brid.gy doesn’t want the pages to be crawled either way? http://brid-gy.appspot.com/robots.txt
#
tantek
Zegnat++
#
Loqi
zegnat has 36 karma in this channel (144 overall)
#
Zegnat
If Archive is still upholding robots.txt, they will not be archiving the pages anyway. Might as well not ping them at all.
leg joined the channel
#
tantek
Zegnat, IIRC, Archive still archives but does not display results for pages blocking via robots
#
sknebel
they certainly are moving in the direction of respecting it less and less. they already ignore it for some crawls of US gov websites
#
tantek
and have for a while
#
tantek
deliberately - for the sake of history
#
aaronpk
gosh deciding what things should look like is always such a hurdle when i'm working on a project
#
Zegnat
sknebel, yeah, I remember we discussed that during IWC. They also will not respect future robots.txt anymore. So if someone takes over the site and puts a disallow everything on there the old site is still visible in the archives, I think someone said.
#
Zegnat
But currently active robots.txt for currently first-seen pages are probably still adhered to.
#
Zegnat
Which would be the case for brid.gy
#
tantek
aaronpk, same. which is why I end up trying to document existing presentations for similar things.
#
dgold
if I have all my mentions in a json-encoded list, how can I then resend them all?
#
aaronpk
you'd have to send a post request to your new webmention endpoint with each source and target URL
#
tantek
dgold, I'd see about getting them first in a standard json-encoded list format, like jf2
#
tantek
then using a tool that consumes jf2 and resends them, or, if one doesn't exist, writing one
#
tantek
that way, you encourage further interop / portability
#
tantek
rather than random json
#
aaronpk
eh, to resend the webmentions, all you really need is the source and target URL, and that doesn't really apply to jf2 or mf2
#
tantek
aaronpk, then you get stuck in the trap of writing custom throwaway code
#
tantek
which is an endless pit
#
aaronpk
there isn't a "standard" for representing the source & target URLs though
#
tantek
oh? bridgy does it with h-entry AFAIK
#
aaronpk
I mean you *can* store the whole source post in some format, HTML, mf2 json, jf2, whatever, but then re-sending those webmentions becomes a harder problem.
#
aaronpk
if you already have the source and target URLs it's easier to just use that
#
tantek
hmm, I thought jf2 had a way to encode those
#
Zegnat
jf2 encodes documents, not mentions. Just double checked.
#
tantek
entries, beyond documents
#
tantek
and a list of mentions is just another list of entries
#
aaronpk
that sounds like vocab
#
tantek
they've got URLs and datetime stamps
#
tantek
no need to reinvent a bunch of stuff
#
Zegnat
Hmm, do mentions have time stamps? I don’t save mine. That’s something I should think about.
#
tantek
aaronpk, it's possible we may need a new term or two
#
Zegnat
And I said documents because that is what jf2 calls their own things, I guess. Sorry for causing confusion :)
#
tantek
yes absolutely, webmentions happen (like any HTTP request) at a particular point in time
#
tantek
zegnat, that's like document like "HTML" vs application like JS
#
tantek
you may be overinferring from "document"
#
Zegnat
https://dissolve.github.io/jf2/#conformance - starts with talking about documents and has a “non-exhaustive list of examples of documents”. Which is why the term was in the front of my mind tantek
#
dgold
wonder if I could use Huginn to automate webmention sending when there are new posts on my site
#
Zegnat
You can encode anything in jf2, as long as you add a custom prefix to the property names to make clear it isn’t an actual microformat you are encoding. But at that point you are basically just using a default JSON struct. So I don’t see the point of encoding webmentions in jf2.
#
Zegnat
tantek, yeah, I realise HTTP requests happen at a specific time. What I am thinking about is if it is in any way important to store that data.
[eddie] joined the channel
#
aaronpk
Zegnat: I store it so I can sort by that, since sometimes there is no published date on the source document
#
sknebel
dgold: yes, I think huginn should be able to follow your feed and run something for each new post. (handling updates might be trickier, depending on how they appear in your feed - that'd be easier if you did it integrated with your build step)
#
tantek
Zegnat, it's important as it's an actual physical event that occurs (electrons and such)
#
tantek
as compared to the *claim* of dt-published
#
tantek
it's not a problem now/yet, but as webmention grows in popularity, there will be dt-published bugs (abuses?) that may be "checked" via such "webmention received" time
#
dgold
sknebel: its an incredibly powerful tool, i've been progressively moving all my ifttt/zapier stuff to it, I just lack tuits
[colinwalker] joined the channel
#
tantek
nice post [colinwalker] !
#
sknebel
dgold: do you run hugo on a VPS?
#
sknebel
I know, I looked at it a few times but never had an important enough reason to go through setting it up
#
dgold
sknebel: yes - that's on an ovh server in the netherlands
#
aaronpk
wow Huginn looks intense
#
tantek.com
edited /Webmention (+25) "REC flow chart"
(view diff)
#
Zegnat
I am not sure why my webmention endpoint will ever care about publishing datetime. But I am definitely going to consider storing webmention receiving time in my next endpoint implementation
#
dgold
I had to move the whole operation to a server in order to utilise micropub
#
tantek
Zegnat, dt-published is just one of the h-entry fields that a webmention receiver will likely want to cache
#
aaronpk
Zegnat: if you ever plan on re-sending old webmentions like we're talking about for importing old ones, the published date would be relevant so that your new webmentions don't all look new
#
tantek
or what you do you mean by "will ever care"?
#
dgold
aaronpk: its _awesome_, once you get your head around liquid
#
aaronpk
it uses Liquid templates for defining events?!
#
dgold
XPath, sorry, don't know why I said liquid
#
Zegnat
There is nothing in the webmention spec about parsing microformats. That’s not something my endpoint does either. So whatever is in dt-published doesn’t matter to me, tantek
#
tantek
Zegnat, not strictly true. if you read the Updates section you will see references to microformats
#
aaronpk
hm I should file an issue to get them to support the payload that my GPS tracker app sends https://github.com/huginn/huginn/wiki/Agent-Types-&-Descriptions#user-location-agent
#
Zegnat
Well, yes, if you extract data from webmentions you should update said data, is what the spec says. https://www.w3.org/TR/webmention/#updating-existing-webmentions
#
Zegnat
But if I do not import dt-published the first time, I also do not have it to update on an update ¯\_(ツ)_/¯
#
dgold
how do you handle webmentions, Zegnat?
#
Zegnat
Currently, extremely crudely. I accept them and store them and send myself a notification that someone mentioned something
#
Zegnat
My endpoint cares about mentions, and the mention being valid. Not about the content of the documents. Which is why updates only make sense to me as far as “should I redo the notification?” And in my flow, I am not sure how much value timestamps have at all.
#
Zegnat
Note that I do not show replies, so my usecase will not match yours, dgold
[asuh] and [tantek] joined the channel
#
asbjor.nu
created /User:Asbjor.nu (+53) "Created page with "[https://asbjor.nu/ Asbjørn Ulsberg, aka. asbjornu].""
(view diff)
#
@sull
@ummjackson good question. been a while since I followed this. i’ve used indie webmention in past too. https://indieweb.org/reply
(twitter.com/_/status/929782848852234240)
[kevinmarks] joined the channel
#
[tantek]
Finally caught up on all the webmention storing / serving discussion
#
[tantek]
I see a lot of these services as nice building blocks to more quickly bootstrap your IndieWeb implementation or setup
#
[tantek]
And then prioritize new features vs rebuilding things you're using services for
#
[tantek]
Aaronpk interesting points / experience about authorship for a feed rather than a post
#
[tantek]
Perhaps it's worth a separate algorithm for feeds to reduce http requests for common cases?
#
aaronpk
likely yes
#
[kevinmarks]
tantek - pass through handler wiht internet archive support is mention-tech
#
aaronpk
probably also worth re-evaluating whether the main /authorship algorithm should have that http request so high up in the list
#
aaronpk
I feel like making another http request should be a last resort
#
Zegnat
Note that expanding authorship to feeds has been discussed and documented: https://indieweb.org/authorship#Questions
#
aaronpk
that's slightly different
#
Zegnat
How so?
#
aaronpk
i'm not talking about determining the author of a feed, but determining an author of each item within a feed
#
Zegnat
Aah, alrighty
#
Zegnat
From tantek’s “authorship for a feed” comment I thought he meant determining the author of a h-feed
#
aaronpk
i've decided I don't care about trying to determine the author of a feed, because multi-author feeds are going to be pretty common in my case
#
[kevinmarks]
is "use the feed author if you can't find the entry author" OK?
#
aaronpk
for Atom and RSS yes
#
[kevinmarks]
for h-feed too though
#
aaronpk
nah you don't need to have an "author" property of an h-feed in order to be able to use an h-card that happens to be in the h-feed
#
[tantek]
"Author for a feed" just means default author for entries in the feed
#
[tantek]
And explicitly why we should NOT have an "author" property on h-feed
#
[tantek]
Aaronpk is right
#
[kevinmarks]
right, but that is authorship
#
[tantek]
Atom and rss got this wrong
#
aaronpk
I do think we need part of the /authorship algorithm that applies to determining authors of items within an h-feed
#
[tantek]
Entries :)
#
[kevinmarks]
I have an author for the feed
#
[tantek]
Don't make me define item ;)
#
aaronpk
in tantek's case, his h-feed is a child of his h-card, so that's a totally different way of looking at it
#
aaronpk
I said "item" because that's what the mf2 parser returns, since they may not all be h-entry
#
Zegnat
I have an author on my feed too, [kevinmarks]. But I recognise far from all feeds do so.
#
[kevinmarks]
yes, feeds have multiple authors, but the single author feed case is common
#
[kevinmarks]
so there is a DRY tension
#
[tantek]
aaronpk++ score one. /me checks to see if his h-feed has a direct h-event child
#
Loqi
aaronpk has 86 karma in this channel (1467 overall)
gRegorLove joined the channel
#
Zegnat
Hmm, 7.1 is the first part where the authorship algo fetches an external page, or am I missing an HTTP request higher up, aaronpk?
#
aaronpk
that's correct
#
Zegnat
I can totally see how adding some other checks before that (e.g. is there an h-card parent object to use) is useful, but I don’t think that’s “so high up in the list”
#
aaronpk
it happens surprisingly often
#
aaronpk
I *think* it would still get good results moving 7.1 to after 7.4
#
[kevinmarks]
moving 7.4 to before 7.1-3 ?
#
aaronpk
yes I guess that would be the way to say it
#
Zegnat
Yeah, I think what [kevinmarks] says. 7.2 and 7.3 seem to require 7.1
#
[kevinmarks]
that seems sensible to me; does someone parse enough pages to make a statistical comparison?
#
Zegnat
I think that could break things on my page.
#
aaronpk
XRay does
#
aaronpk
i'd have to add some sort of logging and let it run for a while tho
#
Zegnat
I have my name in the footer marked up as an h-card, but if that is the only thing you parse you will never find my photo.
tantek joined the channel
#
Zegnat
You must fetch the external h-card to get the full information about me.
#
tantek
is the full information needed for reader use-cases?
#
aaronpk
what i'm looking for in the reader use case is the name and photo
#
Zegnat
Depends if readers want to show photos.
#
Zegnat
In general most implementations seem to want photos from the authorship algo
#
tantek
can be done lazily
#
tantek
if the h-card you have lacks a photo, and you need one, go get it from the h-card's uid/url
#
[kevinmarks]
which would argue for the order change aaron mentions
#
[kevinmarks]
as you can use the local h-card cache first
#
aaronpk
I haven't even begun to think about caching for this yet, but it does seem like a reasonable expectation to cache author info like the photo
#
[kevinmarks]
sorry, I meant the h-card on the page already
#
Zegnat
After the order change, I think you will only get my first name and URL from the algo. But that might be enough for most use cases?
#
aaronpk
I think what I would do in that case Zegnat is fetch your home page to find the photo and cache it, then only fetch it again periodically
#
Zegnat
Mini h-cards are a thing. E.g. all /person-tags on a page are one. Those might resolve in some circumstances as “h-card on the current page with a matching url”.
#
tantek
what is fetch it again periodically?
#
Loqi
It looks like we don't have a page for "fetch it again periodically" yet. Would you like to create it?
#
[kevinmarks]
so display name and url immediately, then go looking for pic, rather than stall?
#
tantek
fetch it again periodically is a pre-WebSub approach to updating information, also known as polling.
#
loqi.me
created /fetch_it_again_periodically (+127) "prompted by tantek and dfn added by tantek"
(view diff)
#
[kevinmarks]
stop trying to make fetch it again periodically happen
#
aaronpk
subscribing to a home page to get notifications of profile photo changes... hmm
#
tantek
WebSub: stop periodically making fetch it again happen
#
Zegnat
Yeah, you would need your person cache to subscribe to all URLs, I guess :P
#
tantek
aaronpk, well yeah WebSub works for subscribing to any page to get updates
#
tantek
zegnat, only the ones you want updates for
#
aaronpk
stop making work for me :P
#
Zegnat
Wouldn’t that be all of the h-card u-urls in your person cache though, tantek? E.g. for watching for photo changes
#
tantek
kevinmarks that's worthy of @w3cmemes2
#
tantek
actually WebSub is still a PR
#
tantek
so go tell your AC reps that you know to vote for it still :)
#
aaronpk
turns out I will be subscribed to a lot of home pages anyway because most of the feeds I follow are home page feeds
#
aaronpk
seems like a good opportunity to update local h-card caches
#
tantek
so you just need to subscribe to Zegnat's separate h-card page
#
tantek
as part of that "infrastructure"
#
Zegnat
I’ll send out a ping whenever I update my weight, I guess :P
#
tantek
kevinmarks, update that meme to "WebSub is now a REC", then have it ready for when WebSub actually *goes* to REC!
#
aaronpk
Zegnat: right now if you use superfeedr's subscription service to subscribe to your home page it will already send websub notifications when your weight is updated
#
tantek
then you can do the blog post about it with that as the featured image :)
#
Zegnat
Ah, interesting, I haven’t looked at superfeedr at all yet
#
aaronpk
it's a bit confusing because it's 3 things
#
tantek
kevinmarks++
#
Loqi
kevinmarks has 6 karma in this channel (287 overall)
#
[kevinmarks]
does loqi have that template?
#
Zegnat
Also note that with mini h-cards, e.g. because they show up inside a post, the display name might not be the name chosen by the person the u-url points too. So you could theoretically end up with a nickname chosen by someone else in your data.
#
[kevinmarks]
right, but it would correct when you do fetch the url
#
[kevinmarks]
so maybe we add an async phase to authorship
#
tantek
BTW the WebSub poll closes TODAY
#
tantek
and then SocialWG has to process the comments
#
[kevinmarks]
how is it looking?
#
Zegnat
That would fix that, yes, [kevinmarks]. So we would move 7.4 up to above 7.1, but have to make clear that the algo cannot stop at that point?
#
tantek
no one is against it which is good. just a couple of minor proposed changes
#
Zegnat
Wouldn’t actually cut out any HTTP GETs then
#
[kevinmarks]
it would defer them, which is good
#
aaronpk
Zegnat: not "cannot", but makes it more clear that it's optional
#
Zegnat
But if you stop you may have just taken in false information not chosen by said author themselves.
#
aaronpk
only if it's on a different domain
#
aaronpk
depends on the consuming use case
#
aaronpk
also you can always ignore the name and photo and use what's in your cache, having used /authorship to just determine the URL
#
[kevinmarks]
this gets trickier if the h-card is on a comment
#
aaronpk
yes like I said, depends on the consuming use case
#
Zegnat
call her.
#
Zegnat
Yeah, that’s the problem with “first h-card” IMHO. I have a post with <a href="http://isabelforester.com" class="h-card">Isi</a>. If that post is ever displayed higher up the feed from anything with u-author isabelforester.com (be it a repost, a comment, whatever) the algo will end with “Isi” as the author name. While in reality that is just how I
#
aaronpk
I think you might be making up theoretical problems
#
Zegnat
Actually, class="u-category h-card" sorry
#
tantek
Zegnat, aaronpk, already answered that with "different domain"
#
tantek
Zegnat, what consuming code have you used that had this problem?
#
tantek
Or is it a problem you think *might* happen in theory?
#
Zegnat
None, because current algo doesn’t have this problem
John__ joined the channel
#
Zegnat
It is a problem that would happen if we accept the first h-card to match u-author on the current page
#
Zegnat
Which I think was the change being proposed that we were discussing?
#
tantek
who made that proposal?!?
#
tantek
in what step of /authorship?
#
Zegnat
Moving 7.4 to before 7.1 in /authorship algo is what was being discussed...
#
Zegnat
Which I believe would have the effect I just described.
#
aaronpk
does /authorship apply to finding author info of comments on an h-entry? I don't think it was intended for that but it's not explicit.
#
Zegnat
It is intended for any h-entry. Comments are also marked up as h-entry
#
Zegnat
Any mini h-cards (e.g. person-tags) on the same page (could be elsewhere in a feed) could be treated as the author’s h-card as long as they are first in source order. These h-cards may be filled with nicknames chosen by someone else. Of course if we codify the limitation of same domain that problem is basically gone.
#
tantek
no it is intended for the permalink of an h-entry
#
tantek
not "any h-entry"
#
Zegnat
Not as it currently reads …
#
Zegnat
you just “start with a particular h-entry”
#
aaronpk
what is authorship?
#
Loqi
authorship is how to determine who the author(s) of a post are https://indieweb.org/authorship
#
[kevinmarks]
i think Zegnat is right, but I don't think it is a big problem
#
Zegnat
I would also find it substantially less useful if the entire algo is limited to permalink pages only.
#
aaronpk
the scope of /authorship should be made more explicit on that page if it really is intended to only apply to a post permalink
#
tantek
has to now re-read and reload all the context past/present of that page
#
[kevinmarks]
having to fetch each h-entry and h-card url is a fork bomb; it is going to have to be async
#
aaronpk
however as written I think it is mostly applicable to finding the author of an h-entry within a feed as well
#
Zegnat
Yeah, I have definitely used it for h-entry (and h-event) within feeds too.
#
aaronpk
i'm not sure it was ever intended to work for a comment of an h-entry though
#
tantek
it was not
#
Zegnat
If said comment is marked up as an h-entry, it has worked fine for me whenever I used it.
#
Zegnat
And I personally think it should work on any h-entry. Whether it is nested under a different h-entry (as comment or as anything else) shouldn’t matter IMHO.
#
[kevinmarks]
if the entry is a repost
#
[kevinmarks]
which is good component boundary design
#
tantek
Zegnat, can you add your use-cases to a "Brainstorming / Use-cases" subsection on /authorship to at least capture them?
#
Zegnat
Though most of the things I just said only really apply if we move step 7.4 up. I think it is fine as a fallback for when 7.1–7.3 fail.
#
Zegnat
As [kevinmarks] said, it is not a very big problem, and I think it is an issue that is acceptable for a fallback value. Just not for the main value.
#
tantek
^^^ that's a great place to add h-feed and WebSub as columns!
#
tantek
and then see which of those "feed readers" support either / both!
#
[kevinmarks]
I think aaron's point about using the 7.4 answer until the 7.1-3 fetch completes is a key one
davidmead joined the channel
#
Zegnat
Yes, using 7.4 temporary if you want to make 7.1–7.3 async also seems fine to me. You are basically using it as a fallback then as well (e.g. if the async takes too long or fails).
#
tantek
also Woodwind should get added to that list
#
Loqi
ok, I added "Add [[Woodwind]] to all the tables in https://en.wikipedia.org/wiki/Comparison_of_feed_aggregators" to the "See Also" section of /to-do
#
loqi.me
edited /to-do (+101) "tantek added "Add [[Woodwind]] to all the tables in https://en.wikipedia.org/wiki/Comparison_of_feed_aggregators" to "See Also""
(view diff)
#
Zegnat
Using 7.4 if hostnames of the object’s u-url and the h-card’s u-url match also seems fine to me, that way you have insurance that the author influences their own h-card on the page
#
tantek
to-do << add h-feed and WebSub to the columns of this table https://en.wikipedia.org/wiki/Comparison_of_feed_aggregators#Web_feed_and_protocol_support, note which feed readers support either or both, and reach out to other OSS in that table requesting support (filing GH issues accordingly, and linking to them on their indieweb wiki pages!)
#
Loqi
ok, I added "add h-feed and WebSub to the columns of this table https://en.wikipedia.org/wiki/Comparison_of_feed_aggregators#Web_feed_and_protocol_support, note which feed readers support either or both, and reach out to other OSS in that table requesting support (filing GH issues accordingly, and linking to them on their indieweb wiki pages!)" to the "See Also" section of /to-do
#
loqi.me
edited /to-do (+335) "tantek added "add h-feed and WebSub to the columns of this table https://en.wikipedia.org/wiki/Comparison_of_feed_aggregators#Web_feed_and_protocol_support, note which feed readers support either or both, and reach out to other OSS in that table requ"
(view diff)
#
Zegnat
So I guess … moving 7.4 up with certain limitations gets a +1 from me, but wholesale moving 7.4 up without limitations a -1?
KartikPrabhu joined the channel
#
aaronpk
I don't think it needs limitations
#
aaronpk
basically there needs to be a global security concern about using information about other domains
#
aaronpk
but that applies to more than just step 7.4
#
tantek
wow while looking for Mozilla and Pubsubhubbub I found this gem! https://blog.mozilla.org/webdev/2010/07/26/registerprotocolhandler-enhancing-the-federated-web/ <-- aaronpk, were you part of this discussion?
#
aaronpk
wow no I dont think so
#
Zegnat
https://indieweb.org/authorship#Algorithm_Design_Notes - he, the exact photo usecase gets mentioned in the design notes as the reason why the HTTP request comes before looking for a small one. So whoever wrote that (tantek?) apparently thought the same thing as I did
#
tantek
Zegnat, a lot of the authorship algorithm came out of real world examples and the last time we had a flurry of indieweb reader development around here, probably around Monocle and/or Woodwind
#
tantek
searches bugzilla for websub / PuSH
#
tantek
notices the Firefox Bookmarks menu item "Subscribe to This Page..." and wonders how it would make use of WebSub.
eli_oat joined the channel
#
tantek
aaronpk is it even possible for a Browser to be a WebSub subscriber?
#
tantek
the WebSub spec does not mention "browser" anywhere
#
tantek
I'm guessing the answer is "no" since a browser can't really provide a "subscriber callback URL"
#
aaronpk
yeah sadly not
#
tantek
perhaps WebSub WebRTC subscriber could be a future thing, so browsers could receive WebSub notifications over WebRTC?
#
aaronpk
oof that doesn't sound like fun
#
aaronpk
every time I try to learn about WebRTC I run away screaming
#
[kevinmarks]
hm. doesn't the notification api need that?
#
[kevinmarks]
so you could bridge websub to notifications
#
aaronpk
yeah I don't quite understand how they got around that
#
tantek
notification API is for user-level notifications
#
aaronpk
I think cause they limited the scope of the notification API to the client-side
#
sknebel
and webPush api relies on a server run by the browser maker
#
aaronpk
ah right Notification API != Push API
#
tantek
would be weird to use that as plumbing, though maybe websub could share plumbing with whatever notifications API is using?
#
sknebel
since browser makers already run that infra, they might be able to use it/extend it to be used for websub
#
aaronpk
yeah I don't think websub would have anything to do with the notification API since websub doesn't talk about displaying anything
#
sknebel
aaronpk: push doesn't either, it goes to a service worker in the normal case (maybe extensions as an an alternative, not sure?) I think?
#
aaronpk
yeah it could be tied to the push API
#
aaronpk
the browser vendor would run the server that websub hubs would deliver to
#
aaronpk
then that payload would be made available to the JS subscriber
#
sknebel
yes. you need some encryption thing for the push api, so you can't just give the push api endpoint to the websub server though I think
#
tantek
or we figure out how to define a hub that does push notifications itself? or a hub that does WebRTC outbound connections to browsers?
#
aaronpk
well WebSub requires the subscriber to be an HTTP endpoint
#
aaronpk
so i'm not sure how to make that work with WebRTC directly
John__ joined the channel
#
[kevinmarks]
push is a better fit
#
sknebel
I'm also not sure I want my browser to talk WebRTC to all my rss feed subscriptions, revealing my online times, IP address, ... in the process, so I wouldn't want that in the hubs
#
aaronpk
yeah I don't think it's good for that use case, but WebSub has other use cases that it might be more appropriate for
#
[kevinmarks]
if you have a service worker, then push can wake it up, which seems very close to a websub callback
#
dgold
someone has a wonderful service to create access tokens via indieauth - what is that called/located?
#
sknebel
what is tools?
#
Loqi
Here are some tools that are helpful for building and debugging an IndieWeb site https://indieweb.org/tools
#
dgold
gimme a token!
#
dgold
I was typing give me a token
#
sknebel
what is gimme a token?
#
Loqi
It looks like we don't have a page for "gimme a token" yet. Would you like to create it?
#
aaronparecki.com
edited /tools (+23) "move indieauth tools to a new section"
(view diff)
#
sknebel
gimme a token is a helper to obtain a an access token from your IndieAuth endpoint: https://gimme-a-token.5eb.nl/
#
loqi.me
created /gimme_a_token (+141) "prompted by sknebel and dfn added by sknebel"
(view diff)
#
kaja.sknebel.net
edited /gimme_a_token (+1) "linkify ('x is y. <url>.' pattern)"
(view diff)
#
sknebel
argh, I need to fix that pattern
#
aaronpk
that looks like it worked
#
tantek
skenebel if you use any social media silo, you are "revealing my online times, IP address, ... in the process" to them
#
www.svenknebel.de
edited /gimme_a_token (-1) "sorry for the noise, my bot is being unhelpful Undo revision 41826 by [[Special:Contributions/Kaja.sknebel.net|Kaja.sknebel.net]] ([[User talk:Kaja.sknebel.net|talk]])"
(view diff)
#
tantek
because of their JS that is constantly updating / pushing updates to you like new posts, comments, red number updates etc.
#
aaronpk
tantek: yes but that's very different than revealing that information to everyone you *follow* on those services
#
tantek
sknebel: even if you "just" use github, you are doing that
#
sknebel
tantek: only if I open them, not by just having their RSS feed subscribed in firefox
#
tantek
aaronpk, who said "everyone"?!?
#
tantek
you are only revealing it to the *hub*
#
sknebel
(unless people broke service worker privacy that much)
#
tantek
not "everyone you follow"
#
tantek
sheesh
#
vanderven.se martijn
edited /authorship (+1673) "/* Other theoretical */ Documenting some theoretical issues with 7.4, that may come to light if 7.4 is moved up in the algo step order. Links to chatlog. Documented here after prompt by tantek."
(view diff)
#
dgold
thanks sknebel
#
aaronpk
given that anyone can run their own hub, I think the concern is valid
#
sknebel
every RSS feed I follow
#
tantek
but not in practice
#
aaronpk
there's a wordpress plugin that is a hub, so I don't think that's far off
#
tantek
did it file an implementation report?
#
tantek
for WebSub?
#
Loqi
[Matthias Pfefferle] Description This plugin is a simple way to let people know in real-time when your blog is updated. PubSubHubbub is widely adopted and is used by Google Reader, Google Alerts and many other services. WebSub provides a common mechanism for communic...
#
tantek
if no implementation report, then yes I think it's far off, because it's not really tested with websub.rocks etc.
#
aaronpk
it's been around for a long time tho, way before websub
#
aaronpk
o.O "Active installations: 100,000+"
#
tantek
will be interesting to see what you see in reader logs for subscribing to hubs
#
aaronpk
looking forward to it
#
loqi.me
created /give_me_a_token (+26) "prompted by sknebel and dfn added by sknebel"
(view diff)
eli_oat2 joined the channel
#
tantek
!tell snarfed does your indiestats project show somewhere # of total comments received/posted, and similarly for likes, reposts, RSVPs?
#
Loqi
Ok, I'll tell them that when I see them next
#
tantek
!tell snarfed even better, a graph over time as the numbers have grown?
#
Loqi
Ok, I'll tell them that when I see them next
leg joined the channel
#
aaronpk
In other news, Monocle as a minimum viable Microsub server is making progress
#
tantek
is Microsub worthy of writing up as a brief W3C Note?
#
aaronpk
Not til there's at least one implementation :-)
#
dgold
when a client sends a mp-syndicate-to, is the value given the uid or the name?
#
aaronpk
isnt that explicit in the spec?
#
dgold
it is, thank you
gRegorLove joined the channel
#
sknebel
aaronpk: btw, since someone asked in Berlin when Websub was mentioned, did you look at any of the APIs hosted readers offer for alternative clients while coming up with it?
#
dgold
oh, wow, thanks KartikPrabhu
#
dgold
just visited my own site and did that trick, my svgs aren't as large as all that, but they certainly need adjustment
#
Zegnat
I think you mean microsub, not websub, there, sknebel
#
sknebel
Zegnat: indeed
#
KartikPrabhu
dgold: yeah, I just finished updating mine too
#
www.svenknebel.de
edited /bookmark (+107) "/* Jeremy Keith */ displays received WMs"
(view diff)
#
dgold
KartikPrabhu: what did you set yours to? 50 seems awfully big
#
KartikPrabhu
dgold: my site logo is 100 and icons are 25
#
Zegnat
It is nice to now have a name for those huge SVGs I see when browsing the web :D
#
Zegnat
KartikPrabhu++ for that link
#
Loqi
kartikprabhu has 6 karma in this channel (156 overall)
#
dgold
KartikPrabhu++ agreed
#
Loqi
kartikprabhu has 7 karma in this channel (157 overall)
#
sknebel
sounds like someone should put it in a browser extension for ppl that block CSS ;)
#
KartikPrabhu
sknebel: a browser extension for simple progressive enhancement is overkill no?
#
dgold
huh. the biggest offender on my site is the bigfoot.js for footnotes
#
dgold
can't see a way to repair that
#
KartikPrabhu
dgold: maybe send a PR to bigfoot
#
Zegnat
Your footnotes need SVG?
#
Zegnat
I use https://www.sitepoint.com/accessible-footnotes-css/ for mine, but I guess I am super not-fancy with them
#
dgold
Zegnat: bigfoot.js is a thing to have inline feetneet
#
dgold
KartikPrabhu: will do
#
KartikPrabhu
dgold: framing it as a minor progressive enhancement mighthelp
#
Zegnat
Might have to check out bigfoot then. I could use on making my footnotes a little nicer.
#
dgold
i started using bigfoot on the recommendation of an accessibility user
#
dgold
not sight-limited, but mobililty impaired. I guess I didn't fully consider the implications
#
Zegnat
The accessible footnotes thing seems nice for mobility impaired. You do not need pinpoint precision for clicking as the entire word box around the small footnote marker is click-able.
#
Zegnat
Maybe I should make that more clear through styling.
#
dgold
Zegnat: but such a small word-box is inherently difficult to hit
#
dgold
the bigfoot gives a big fat (...) to hit
#
KartikPrabhu
I have similar issues with my marginalia buttons. I couldn't use them on mobile despite not being impaired in any way
#
KartikPrabhu
not sure how to fix it
#
Loqi
[Daniel Goldsmith] Tracks in the Dust
#
Zegnat
enabled scripts
#
Zegnat
enables css quickly afterwards because huge SVGs
#
dgold
Zegnat: I haven't pulled the fix yet!!!
#
KartikPrabhu
fatwigoo--
#
Loqi
fatwigoo has -1 karma
barpthewire joined the channel
#
Zegnat
Actually the entire text “mostly asexual[1]” is the clickable area, you just don’t see it
#
Zegnat
I should fix my styling to make for bigger clickable areas. See first paragraph here for my footnote: http://vanderven.se/martijn/gender/
#
dgold
that didn't work...
#
Loqi
definitely
#
dgold
i dropped the width: and height: to 32 for the svgs, they still show as GIANT MASSIVE IMAGES
#
KartikPrabhu
Zegnat: what happens if there is a footnote next to a link?
#
KartikPrabhu
dgold: example URL?
#
sknebel
Zegnat: would be nice if at least the hover style reflected that it is a link - not even the mouse pointer changes right now
#
Zegnat
KartikPrabhu, I can’t have footnotes next to links, that’s a limitation of the system. The superscript number is added by CSS so the actual footnote link is always a piece of text
#
KartikPrabhu
Zegnat: aah I see. Nice use of CSS counters though!
#
Zegnat
I take 0 credit, all from the sitepoint article I linked. Heydon also worked on that to test the accessibility.
#
Zegnat
sknebel, yeah, current CSS is specifically made so it looks like only the superscript number is interactable. I am now considering changing that for people with mobility impairments. Was a good point by dgold.
davidmead joined the channel
#
sknebel
(also, touch device users are also "impaired" at hitting small targets)
#
Zegnat
Yes, but that kinda fixes itself. If they try to touch my superscript chances are they hit because there is an entire “normal” link box there
#
sknebel
(and for finger-touch devices, hover isn't really a thing, so ideally the non-hover style shows it as well)
#
KartikPrabhu
I think there is some CSS proposal to have a "media query" for pointer precision or something
#
Zegnat
I am just hesitant to add a style that could be confused with the word being an off-page link
#
sknebel
on a pen device, I might spend unnecessary time aiming on a small target I can hit - but a hover style would work for many of those (since most pen devices track while hovering as well)
#
sknebel
I wonder if you could make a (dotted?) line that "connects" to the number with CSS somehow
#
KartikPrabhu
sknebel: with border bottom and right possibly
#
Zegnat
Oh, sknebel, before I forget: http://www.metadataworkinggroup.com/specs/ (that WG is Adobe, Apple, Canon, Microsoft, Nokia, and Sony) is the spec that has an XMP extension for tagging people’s faces with their names within photo metadata :)
#
Zegnat
With that I am off for bed. I will consider trying to hook the number into the link to make the area more clear for footnote links. Thanks for the ideas all!
#
sknebel
image_metadata << http://www.metadataworkinggroup.com/specs/ (that WG is Adobe, Apple, Canon, Microsoft, Nokia, and Sony) is the spec that has an XMP extension for tagging people’s faces with their names within photo metadata
#
Loqi
ok, I added "http://www.metadataworkinggroup.com/specs/ (that WG is Adobe, Apple, Canon, Microsoft, Nokia, and Sony) is the spec that has an XMP extension for tagging people’s faces with their names within photo metadata" to the "See Also" section of /image_metadata
#
loqi.me
edited /image_metadata (+212) "sknebel added "http://www.metadataworkinggroup.com/specs/ (that WG is Adobe, Apple, Canon, Microsoft, Nokia, and Sony) is the spec that has an XMP extension for tagging people’s faces with their names within photo metadata" to "See Also""
(view diff)
#
sknebel
Zegnat: thanks and good night!
#
KartikPrabhu
wow! the spec guidelines are a PDF! come on!
#
gRegorLove
what is image metadata?
#
Loqi
image metadata is information captured along with an image beyond the actual pixel data https://indieweb.org/image_metadata
snarfed joined the channel
#
Loqi
snarfed: tantek left you a message 1 hour, 46 minutes ago: does your indiestats project show somewhere # of total comments received/posted, and similarly for likes, reposts, RSVPs?
#
Loqi
snarfed: tantek left you a message 1 hour, 45 minutes ago: even better, a graph over time as the numbers have grown?
#
snarfed
!tell tantek yes! indie map does count responses by type, both overall and per site. http://www.indiemap.org/docs.html#stats . see the "mf2 class frequency" and "mf2 class frequency by site" graphs
#
Loqi
Ok, I'll tell them that when I see them next
#
snarfed
!tell tantek not over time though, since the crawling is a big chunk of the total work.
#
Loqi
Ok, I'll tell them that when I see them next
#
snarfed
!tell kevinmarks,zegnat btw http://www.indiemap.org/ is a good source for the kinds of stats you all were idly guessing about while debating authorship and representative h-card. easy to data mine w/SQL. happy to help!
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
Indie Map is a public IndieWeb social graph and dataset. 2300 sites, 5.7M pages, 380GB HTML + mf2. Social graph API and interactive ma...
#
snarfed
aaronpk++ nice!
#
Loqi
aaronpk has 87 karma in this channel (1468 overall)
KartikPrabhu joined the channel
#
aaronpk
I think I got all the UI stuff done. next I can focus on hooking up all the plumbing
eli_oat joined the channel
#
@ubiquitic
Reading about #webmentions… the 2d “famous” post I stumble across is an ode to spam: https://adactio.com/journal/6469 https://pbs.twimg.com/media/DOeD-vbVwAEAjMu.jpg
(twitter.com/_/status/929856968189140992)
davidmead and tantek joined the channel