#indiewebcamp 2015-09-30

2015-09-30 UTC
tantek, parzzix, snarfed, lukebrooker, emceeaich, wolftune and nitot joined the channel
#
tantek.com
edited /post-type-discovery (+614) "Normative References h-entry for property names in the algorithm, Activity Streams for JSON, microformats2-parsing for canonical microformats2 JSON, cite normative references inline where needed"
(view diff)
#
acegiak
Anyone else using bridgy finding that their POSSE tweets are coming up as mentions?
#
GWG
acegiak: I've heard some things from snarfed about unintended consequences of recent upgrades
#
GWG
acegiak: Lots going on.
#
tantek
I wonder if it might be possible to put some of the response-type discovery smarts on the webmention endpoint itself rather than depending on Bridgy to get the typing right
#
tantek
as far as mention vs reply at least
#
GWG
acegiak: If it makes you feel better, snarfed is thinking of you. https://github.com/pfefferle/wordpress-semantic-linkbacks/issues/40
#
tantek
obv fave & repost need to be natively determined / sent by Bridgy
#
aaronpk
it's more a question of getting the bridgy markup from the silo right
#
aaronpk
but that text description is coming from bridgy and it doesn't have to
#
GWG
tantek: Isn't this the problem that dates back to pingbacks? How to show context appropriately?
#
aaronpk
it acts as a nice fallback if the webmention endpoint doesn't do anything other than parse plain h-entrys
emmak joined the channel
#
tantek
sure - I like the idea of simplifying proxies though
#
tantek
it makes the individual sites more robust
#
acegiak
I've been a little absent so I'm not 100% up to date on what's happening
#
acegiak
but I'm aware we need to work out a way to make salmentions less hacky so it can go in the main semantic linkbacks repo
#
GWG
acegiak: Bridgy has been through a round of enhancements that have prompted some issues. I haven't gotten many mentions of late, as I've been less active of late, due injury.
#
GWG
But let's see...on the WordPress side, a bunch of pushes on Semantic Linkbacks and Webmentions...nothing earth shattering.
#
GWG
I'd been rewriting Post Kinds.
#
acegiak
GWG: you always are :P
#
GWG
acegiak: I'm trying to get it perfect.
#
GWG
acegiak: Either way, I keep thinking of ways to make it better or just designed better.
#
GWG
acegiak: So, I was teaching it to work with mf2 properties, even if they aren't in the post editor, as a way of iterating faster
#
GWG
So, a Micropub client can actually add functionality without creating a WordPress based Post UI yet. More parsing from the incoming URL...etc.
#
GWG
acegiak: So, now, just to annoy you, the plugin will look in either the in-reply-to property or the cite property for the URL, for example.
#
acegiak
I'll just wait for the update and see if I need to fix anything :P
#
GWG
I have been waylaid by a medical problem. Limits my typing time
#
acegiak
that sucks
#
Loqi
lolz
#
acegiak
I've just been super busy and then super stressed about not being busy :P
#
GWG
I agree. I have to pace myself.
#
GWG
Also did some contributions to WordPress Core, or tried to.
#
GWG
And I have to work on a fix for Syndication Links based on an opened issue.
#
tantek
Loqi, I think your recent lolz have been inappropriate.
#
GWG
And I owe snarfed unit tests for Micropub
#
tantek
gives Loqi an ounce of empathy.
#
Loqi
grabs the ounce of empathy.
#
GWG
acegiak: It is good to have you back though. You are the most engaged WordPress Indieweb user out there.
#
GWG
tantek: By the way, attempts to get hfeed fixed in the new twentysixteen WordPress theme failed. But I got them to stop styling hentry.
#
aaronpk
I am attempting to implement the post type discovery algorithm and use it to sort my posts into categories now
#
aaronpk
will report back soon
emceeaich joined the channel
#
aaronpk
hmm "If the post has no content property, or has no name property..." but what is the post if there are neither?
#
aaronpk
I guess I kind of assumed a note post would have to have content of some kind
#
KartikPrabhu
there always is an implied name no?
#
KevinMarks
could just be a like?
#
aaronpk
then it's a "like" post
#
emmak
aaronpk: when sending the "syndication" property in a micropub create request, should it use array syntax?
#
emmak
do you know what ownyourgram does in this case?
#
aaronpk
emmak: the syntax allows any property to be sent as a string or array
#
aaronpk
for properties that can have multiple values, it's fine to send it as a string if you're just setting a single value
#
aaronpk
ownyourgram sends a string for syndication, but an array for category
#
KartikPrabhu
even if only one category is set?
#
aaronpk
pretty sure
#
KartikPrabhu
would be good to conform to mf2 and send all props as arrays?
#
aaronpk
no that's just awkward
#
emmak
so micropub endpoints should be able to understand both, for any property that could be an array?
#
aaronpk
emmak: yeah
#
KartikPrabhu
if micropub is intended as a "reverse mf2" then it might want to send all props as arrays
#
aaronpk
but micropub also wants to be easy for people to use, and the simplicity of prop=value is important
#
aaronpk
so i'm reading through https://indiewebcamp.com/post-type-discovery#Algorithm and i'm pretty sure it's possible to get to step 12 with a post that has no content property and an empty string for name, and no other properties
#
KartikPrabhu
but yhen it off loads deciding which prop can be an array and which is only a string to the user anyway
#
aaronpk
KartikPrabhu: if you wanted you could always store everything as an array, and just change strings to arrays in the micropub request
#
KartikPrabhu
aaronpk: yeah seems correct
#
aaronpk
my new micropub endpoint actually does a hybrid of that, when it goes to store the post in microformat syntax it changes any value to a string if there's only one value for that property
#
KartikPrabhu
aaronpk: I have been laggin behind on micropub developments with my endpoint so might have write some code to see how easy/hard it is
#
KevinMarks
aaronpk: if you get to 11 or 12 and name content are both empty, does it matter what kind of post it is?
#
KartikPrabhu
aaronpk: is step 11 supposed to give False if name is blank string an there does not exist a content?
#
aaronpk
KevinMarks: i'm not sure, i guess i might want to not show a post if there is no content i know how to show, and the result of the algorithm as it stands won't tell me that
#
aaronpk
KartikPrabhu: i don't think so, the algorithm is written to stop processing only if the step says "it is a ___ post"
#
KartikPrabhu
aaronpk: oh. if the post has no content property it immediately satisfies Step 6 and is a note
#
KevinMarks
what if it has a summary?
#
aaronpk
summary doesn't affect post type according to this
#
notenoughneon.com
edited /Micropub (+161) "/* Create */ clarify array properties with single value"
(view diff)
#
aaronparecki.com
edited /Micropub (+22) "/* Create */ add full example syndication URL"
(view diff)
#
KevinMarks
right - I'm thinking of the case where the forntpage has summary not content, but does haave a title
#
KevinMarks
so this will see them as notes, when they are articles
#
KartikPrabhu
aaronpk: write so if a post has "no content property" it goes to "it is a note post" in Step 6
#
aaronpk
k I will document that
#
KevinMarks
hm, and http://aaronparecki.com/articles would all show up as notes as they just have titles
#
KevinMarks
s/title/name/
#
Loqi
KevinMarks meant to say: hm, and http://aaronparecki.com/articles would all show up as notes as they just have names
#
aaronpk
some of them have photos so they would actually be classified as photo posts
#
KevinMarks
actually, they all have photos, except lots of them are http://aaronparecki.com/images/article-icon.png
#
aaronparecki.com
edited /post-type-discovery (+588) "/* Issues */"
(view diff)
#
aaronpk
oh that's my fault, the photo should have been the featured photo
#
aaronpk
actually no, I should be using "featured" for that, lol
#
aaronpk
that seems worth pushing a change to my old site for
nitot joined the channel
#
aaronpk
k fixed
#
aaronparecki.com
edited /post-type-discovery (+217) "/* Issues */"
(view diff)
#
aaronparecki.com
edited /Micropub (+17) "/* Create */ nowiki"
(view diff)
lukebrooker and tantek joined the channel
#
aaronparecki.com
edited /post-type-discovery (+351) "/* Issues */ videos"
(view diff)
#
aaronpk
tantek: collecting some issues for you!
#
aaronpk
so far I am liking the method of first determining the post type, then deciding which feeds to put it in
Geng joined the channel
#
tantek
oh that's good feedback too!
lukebrooker joined the channel
#
aaronpk
also realizing there are going to be a lot of post types that are not standardized/formalized that I'm going to want to publish immediately that I'gm oing to have to figure out how to deal with
#
aaronpk
considering just scoping everything under p3k-* properties to avoid the issue (and probably not include that markup on the published post_
#
tantek
true - I should include an FAQ about that - about why it's good that a "note" is the ultimate fallback
#
tantek
plain text first design and all that
#
aaronpk
yep, plaintext first design means note is a good fallback. I'm just not certain an empty note is a good fallback
#
tantek
yes - that's how / why need to use p-summary
#
tantek
and all your new post types should include the plain text equivalent in the p-summary
#
aaronpk
there's still the case of what do I do if there is no or empty content, summary and name
#
aaronpk
oh yes! let me make a note so I don't forget to do that
#
aaronpk
interesting, so I think having this algorithm greatly simplified my code and I am now done with handling collections!
#
tantek
whoa - that's great!
#
tantek
even in it's imperfect state?
#
aaronpk
which was really starting to stress me out
#
tantek
s/it's/its
#
aaronpk
yes but it also establishes a pattern I can easily extend when I start filling up my site with experimental post types
#
tantek
wow I thought I had expressed that pattern on the /posts page already but perhaps not explicitly enough to be clear!
#
aaronpk
which pattern?
#
aaronpk
oh there are more post types here
#
aaronpk
invitation and tag-of posts
#
tantek.com
edited /post-type-discovery (+175) "/* Algorithm */ use summary property if no content property, per issue from aaronpk re: summary"
(view diff)
#
tantek.com
edited /post-type-discovery (+163) "/* Issues */ resolved issue lack of content should use summary as fallback"
(view diff)
#
tantek
yes exactly
#
tantek.com
edited /post-type-discovery (+30) "/* Other Types Under Consideration */ add more types under consideration as pointed out by aaronpk. invitation, tag-of"
(view diff)
#
tantek
aaronpk: could you check resolution to first issue you filed today? https://indiewebcamp.com/post-type-discovery#Resolved
#
aaronpk
hmm well now step 6 does not result in ending the algorithm, is that intentional?
#
tantek
yes - it was easier to use a variable "the content"
#
aaronpk
and now the algorithm is a combination of determining post type and also specifying which value to use, but use for what?
#
tantek
only for determining post type
#
aaronpk
I am not clear what post type is the result of step 6 now
#
tantek
there is no post type as the result of step 6
#
aaronpk
keeps reading
#
tantek
note that 7 and 8 are "Else" clauses
#
KartikPrabhu
might be good to put if else into nested steps
#
aaronpk
so now step 9 ends in the post being a note if there is either content or summary but no name?
#
tantek
nesting-- :P
#
Loqi
nesting has -1 karma
#
aaronpk
i guess this is harder to translate to code now because I don't know what to do on step 6 with the content
#
tantek
aaronpk yes! that's how it worked before but only with if there was content but no name
#
aaronpk
before it had an explicit "then it is a note" which means I can return there
#
aaronpk
tries to re-read this as if he had not seen it earlier
#
tantek
that "then it is a note" is still in step 9
#
tantek
here I'll line break it for you
#
aaronpk
steps 6-8 are now in my code as if(content) { content = content } else if(summary) { content = summary } else { return 'note' }
#
aaronpk
is that what you intended?
#
aaronpk
okay i think this works, but it is slightly less obvious what is happening since I have to carry over the content from step 6-7 all the way to step 13
wolftune joined the channel
#
tantek.com
edited /post-type-discovery (+81) "/* Algorithm */ add line-breaks to make ORs and THENs clearer"
(view diff)
#
aaronpk
yes but the readability was not my confusion
#
tantek
well at least it's more readable
#
aaronpk
could you add this before step 6? "If the post has no "content" property and no "summary" property, or if both are empty, then it is a note post"
#
tantek
that's basically step 6,7,8
#
tantek
"8. Else it is a note post. "
#
aaronpk
right, but it means less carrying state between steps
#
aaronpk
that's what I liked about how it was written originally
#
tantek
no there's no state there
#
aaronpk
mental state
#
aaronpk
right now I read step 6 and I can't actually do anything until I get to step 8, and then I still have to remember what was happening in step 6 to continue
#
kylewm
tantek: should I add my note to the wiki about the content.contains(name) check that i do?
#
kylewm
sorry name.contains(content)
#
kylewm
my check is the opposite of what's there currently
#
aaronpk
kylewm: add that with examples to "issues" and we can get there later :)
#
aaronpk
tantek: yeah i think if you add what I wrote, then you can move the current steps 6+7 down to just before you need them which is at step 13
#
tantek
aaronpk: hmm, the alternative appears to be doing the tests mutiple times
#
tantek
that seems worse
#
aaronpk
oh, hmm
#
aaronpk
true, there will be always an extra check for summary or content
#
tantek
except as it's written now
#
kylewm.com
edited /post-type-discovery (+363) "/* Issues */ note name.contains(content)"
(view diff)
#
aaronpk
right. let me try something with my code...
#
tantek
aaronpk re: your next issue "but no content or summary"
#
aaronpk
hang on one sec
#
aaronpk
I think it might actulally be ok to reorder those
#
aaronpk
i'm writing this code imagining the new sequence of checking content+summary and then later using the values
#
aaronpk
it's a pretty obvious optimization to me that I will need content and summary later and that I can just cache them when I fetch them at first
#
@kylewmahan
realizing tonight that a lot of web hosting companies are basically internet scams preying on inexperienced ppl. big challenge for #indieweb
(twitter.com/_/status/649078256897953793)
#
tantek
yikes kylewm
#
tantek
can we document which ones with citations at /web_hosting ?
#
kylewm
i can't yet
#
kylewm
going to #indiechat :/
#
aaronpk
okay not sure if this will be helpful for you tantek, but I have three versions of this code if you want to see it
nitot joined the channel
#
aaronpk
tantek: v1 is the algorithm as its currently on the wiki https://gist.github.com/aaronpk/f8711b886011033d7dda
#
aaronpk
v2 is with the reordered steps like I described, and you can see i'm fetching content and summary twice like you said https://gist.github.com/aaronpk/9951a0221a878ba7a917
#
aaronpk
v3 is with the simple caching trick so I don't have to go fetch content and summary when I need them later https://gist.github.com/aaronpk/400d35952dbe768fcb66
lukebrooker joined the channel
#
aaronpk
(and yes you should basically be reading this as pseudocode since I'm using functions that are not defined in the gist, primarily $this->get)
#
tantek.com
edited /post-type-discovery (+434) "resolved issue, "unknown" is worse than "note" as fallback, display/presentation is outside the scope of this specification"
(view diff)
parzzix joined the channel
#
tantek.com
edited /video (+70) "/* Shane Becker */ fix links, note latest post was 2013"
(view diff)
snarfed joined the channel
#
tantek.com
edited /post-type-discovery (+396) "resolve issue, reject video for now due to only one "current" u-video example. all other video examples are old or don't use u-video"
(view diff)
#
tantek
sorry to reject video aaronpk - evidene not there for it yet :/
#
tantek
s/evidene/evidence
#
aaronpk
wow i thought there were more!
#
Loqi
tantek meant to say: sorry to reject video aaronpk - evidence not there for it yet :/
#
tantek
aaronpk: well if there are, they're not documented
#
tantek
for example, you're not one yourself!
#
tantek
(you have animated gifs, but no video AFAIK)
#
aaronpk
no it's on there
#
aaronpk
"MP4 recording of laptop screen demonstrating original-post-discovery" does that not count as a video?
#
tantek
you're not using u-video
#
aaronpk
oh I see what happened... that was an autolinked one
#
tantek
plus that would only make 2. I deliberately put the bar at 3 for post type discovery
#
aaronpk
i have actual u-video posts now
#
tantek
and I should add "recent" or "active"
#
tantek
then those should be at https://indiewebcamp.com/video#Aaron_Parecki with the non u-video posts perhaps just as historical addendum
#
tantek
for when you started posting video-like things
#
tantek.com
edited /post-type-discovery (+22) "/* Methodology */ recent (< 1 year old) permalinks"
(view diff)
#
tantek.com
edited /post-type-discovery (+259) "/* Issues */ respond to remaining issues with questions"
(view diff)
#
aaronparecki.com
edited /video (+73) "/* Aaron Parecki */ update video support"
(view diff)
#
aaronpk
huh, ben_thatmustbeme's video URL is actually the HTML page and not the media file
#
tantek
that's even worse!
#
tantek
please document that
#
tantek
!tell kylewm what do you think of making the note / article distinction an "OR" e.g. "name.contains(content) OR content.startswith(name)" means post is a note?
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
Ok, I'll tell them that when I see them next
#
aaronparecki.com
edited /video (+106) "/* Ben Roberts */"
(view diff)
#
tantek
yeah we're pretty far from any kind of video interop
#
aaronpk
yeah apparently
#
tantek
benwerd - do you have any other /video posts
loic_m joined the channel
#
aaronpk
related, Vimeo said there is no way they would link to the raw video file either
#
aaronpk
they wanted to have a way for readers/consumers to get the "embed code" (the iframe) to render in a reader or such
#
aaronpk
so yeah, likely we are far from a good interop story for video
#
tantek
I wonder if they would support some sort of video playback status/controller API through postmessage and their iframe
#
tantek
so you could play pause restart a video via JS controls on your own site
#
aaronpk
interesting
#
aaronpk
they do!
#
tantek
could enable Thisismyjam style home page / reader of other people's jams
#
benwerd
tantek: Just seen this. Not yet. But I will again.
#
tantek
benwerd, it seems we /video posts took a bit of a setback since 2013
#
tantek
video is (still) hard
#
aaronpk
speaking of which, I just dug up some old videos I made in like 2002-2004 and they don't play on my mac because codecs
#
aaronpk
at least that part has gotten better
#
tantek
ironically, older codecs should go patent-free sooner due to expiring, and thus be supported by open source, etc.
#
aaronpk
ha good point
#
aaronpk
i think the next big challenge for me with my new site is figuring out how to handle storing comments and such
#
aaronpk
I think I have too many intermediate steps with my current approach, plus I've encountered several cases where I can't store some URLs
#
aaronpk
still not sure on this though. i'm considering doing something similar to https://github.com/aaronpk/ca3db for storing the HTML of external posts
#
KevinMarks
pestagram has video posts
#
tantek
you know where and how to document that ;)
#
aaronpk
ownyourgram *sends* video posts... I wonder if anyone is handling those
#
KevinMarks
hm, though we don;t seem to see class="u-video" on a <video> tag
#
KevinMarks
oh, hang on it is there
#
KevinMarks
I missed it in the comments
#
KevinMarks
do I file pestagram under silo?
lukebrooker joined the channel
#
kevinmarks.com
edited /video (+208) "/* add Pestagram */"
(view diff)
#
aaronpk
pestagram isn't a silo because it doesn't actually store any content of its own
#
KartikPrabhu
Silo Indieweb Pipe Service = SIPS
parzzix and nitot joined the channel
#
kevinmarks.com
edited /video (+193) "/* unmung */"
(view diff)
#
KevinMarks
there we go, lots of video piped into h-feeds for you subscribing pleasure
#
KevinMarks
aaronpk: BTW, Quill editor won't accept an mp4 URL for video, only a silo link.
#
aaronpk
yeah, it says that tho so it's not a bug ;)
#
KevinMarks
hm, I shoudl get the interneta rchive to mqrk up with h-entry and u-video
Jihaisse and joskar joined the channel
#
voxpelli
aaronpk: I should be handling videos from ownyourgram
interactivist, fkooman and petermolnareu joined the channel
KevinMarks, Geng, friedcell, mlncn, interactivist, elf-pavlik, ttepasse and nitot joined the channel
#
ben_thatmustbeme
!tell aaronpk, tantek re: video tag, my u-video does NOT link to the post, look at the source and see for yourself, pin13 reads it that way (bug?) (unmung parses that differently). It needs to be moved to the <source> tag is all. Should check that before you comment on others implementation.... i mean really, that would make no sense. honestly, i think working out some way for the tag to be on the <video> / <audio>
#
ben_thatmustbeme
/ <photo> tag would be better for when there are multiple sources
#
Loqi
Ok, I'll tell them that when I see them next
Jihaisse1 joined the channel
#
ben.thatmustbe.me
edited /video (+9) "/* Ben Roberts */ correct inaccurate commentary"
(view diff)
#
ben.thatmustbe.me
edited /video (+266) "/* How */ show how to mark up multiple source videos"
(view diff)
LanceyWork joined the channel
#
rhiaro.co.uk
edited /post-type-discovery (+87) "/* Next steps */"
(view diff)
fiatjaf joined the channel
#
fiatjaf
I can't get a correct bridgy URL for any of these comments: https://brid-gy.appspot.com/post/facebook/987100564687701/996189387112152
#
fiatjaf
what am I doing wrong?
nitot, cmhobbs and friedcell joined the channel
#
Loqi
Ok, I'll tell them that when I see them next
[tantek] joined the channel
#
[tantek]
Ben, I don't think I made that wiki edit.
#
[tantek]
However, thanks for the "how to" improvement on /video !
#
Loqi
[tantek]: ben_thatmustbeme left you a message 1 hour, 52 minutes ago: re: video tag, my u-video does NOT link to the post, look at the source and see for yourself, pin13 reads it that way (bug?) (unmung parses that differently). It needs to be moved to the <source> tag is all. Should check that before you comment on others implementation.... i mean really, that would make no sense. honestly, i think working out some way for the tag to be on the <video> / <audio> http://indiewebcamp.com/irc/2015-09-30/line/1443610171345
#
[tantek]
I think perhaps that was aaronpk'so observation? Not sure if he was using pin13 or what. I just did view source at the time I checked, and then cited your example (as the only recent right one!) in post type discovery.
wolftune, rosetree, scoates, LanceyWork and [aaronpk] joined the channel
#
[aaronpk]
Aha my bad
#
Loqi
[aaronpk]: ben_thatmustbeme left you a message 3 hours, 36 minutes ago: re: video tag, my u-video does NOT link to the post, look at the source and see for yourself, pin13 reads it that way (bug?) (unmung parses that differently). It needs to be moved to the <source> tag is all. Should check that before you comment on others implementation.... i mean really, that would make no sense. honestly, i think working out some way for the tag to be on the <video> / <audio> http://indiewebcamp.com/irc/2015-09-30/line/1443610171345
#
[aaronpk]
I put too much faith in mf2 parsing
LanceyWork joined the channel
#
[aaronpk]
I definitely was looking at the parsed result from pin13. I pretty much never actually view source anymore since people's html tends to be harder to read than the mf2 JSON
loic_m and petermolnar joined the channel
#
kylewm
fiatjaf: facebook comment ids aren't very predictable... they vary between ~3 different formats. in this case they want postid_commentid, so the url is https://brid-gy.appspot.com/comment/facebook/987100564687701/996189387112152/996189387112152_996193660445058
#
Loqi
kylewm: tantek left you a message 9 hours, 28 minutes ago: what do you think of making the note / article distinction an "OR" e.g. "name.contains(content) OR content.startswith(name)" means post is a note? http://indiewebcamp.com/irc/2015-09-29/line/1443590514605
#
Loqi
kylewm: tantek left you a message 9 hours, 28 minutes ago: also noted inline https://indiewebcamp.com/post-type-discovery#Issues http://indiewebcamp.com/irc/2015-09-29/line/1443590532732
loic_m_, mlncn, nitot, snarfed and loic_m joined the channel
#
KevinMarks
hm, sounds like we need more tests for u-video markup given source issues
#
aaronpk
also looks like the php parser needs updating for video tags
#
fiatjaf
kylewm: so that's a thing of the facebook API, then?
#
fiatjaf
thank you very much.
parzzix and shiflett joined the channel
#
kylewm.com
edited /post-type-discovery (+509) "/* Issues */ add example of name.contains(content), ask if there are examples of content.startswith(name)"
(view diff)
snarfed1 joined the channel
#
kylewm
fiatjaf: yeah unfortunately. the only reliable way I know of to find the comment ids is to look at the post's /comment edge in the api
#
aaronparecki.com
edited /post-type-discovery (+309) "/* Issues */"
(view diff)
snarfed and elf-pavlik joined the channel
#
fiatjaf.alhur.es
edited /Webmention (+177) "Added went as a "Handling" library"
(view diff)
wolftune, tantek and Geng joined the channel
#
Jeena
No HWC next week either in Gothenburg because I'm in Berlin for the Qt World Summit.
parzzix joined the channel
#
tantek
Jeena - can you make a note in the /next-hwc page? (don't remove Gothenburg - just make a note saying exactly what you said)
gRegorLove and modem joined the channel
#
jeena.net
edited /events/2015-10-07-homebrew-website-club (+84) "/* Where */ Göteborg"
(view diff)
#
Jeena
oh weird formatting
#
tantek
so about PuSH, and it specifies SHA1 which is obsolete, and perhaps specifying any particular crypto is problematic for this reasons, though not doing so may hurt interop
#
tantek
and about the ability to force at PuSH hub to use "none" as a security vulnerability that should be documented in a Security & Privacy considerations section
#
tantek
aaronpk, voxpelli
#
aaronpk
github issue for SHA1 is filed
gavinc joined the channel
#
voxpelli
If there's a from header in PuSH then it's probably in relation to Blaine Cook's ideas I linked to here a couple of weeks/months ago
#
aaronpk
is anyone actually using the From header for anything? if not, I would suggest removing that from the spec
#
tantek
sounds like another good github issue
#
voxpelli
aaronpk: the signature is "sha1=abc123" no? So already pluggable
#
aaronpk
especially since it's totally irrelevant to the actual functionality of PuSH "In the context of social web applications, it is considered good practice to include a From..."
#
tantek
and yeah - if it's not obvious why a feature is there (documented use-case) then lets push to remove it
#
voxpelli
From-headers must be emails according to its spec and would therefore require WebFinger:esque identifiers
#
aaronpk
it just seems that "it is considered good practice" is not very compelling
#
voxpelli
aaronpk: remember our discussion around PuSH based private messaging? That's kind of the purpose of such a header I think
#
tantek
perhaps anywhere you see "it is considered" should be a red flag
#
tantek
voxpelli: but did anyone ever build it and get it working cross-implementation?
#
aaronpk
I would want that to be much more well thoguht out for the private messaging use case
#
tantek
sounds aspirational
#
voxpelli
tantek: everything went through a JSON:pocalypse instead I think, but Blaine Cook is really the one to know about it
#
voxpelli
I think this is the related blog
#
tantek
voxpelli: yeah, first the JSONpocalypse, and now everything JSON-related in W3C is going through a JSONLDpocalypse.
#
voxpelli
tantek: isn't everything on the IndieWeb-wiki just patterns considered to be best practices rather than strict standards that requires compliance? ;)
#
aaronpk
the IWC wiki doesnt make unsubstantiated claims that something "is considered best practice"
#
tantek
I hope not!
#
tantek
voxpelli: there is a big difference between "Here are real world examples with permalinks and real people citations" vs. "it is considered to be" zero citations zero mention of specific people.
#
tantek
AKA "weasel wording"
#
tantek
(not my term, from wikipedia)
#
voxpelli
Know Blaine talked a lot about From headers at Federated Social Web Summit in Berlin 2011, that's when OStatus was about to go JSON and lots of talks about private messaging was in the air
#
tantek
What is it is considered to be
#
Loqi
It looks like we don't have a page for "it is considered to be" yet. Would you like to create it? http://indiewebcamp.com/s/103i
gavinc joined the channel
#
voxpelli
And PuSH 0.4 was the main missing piece in having OStatus go full JSON so it was probably authored around that time
#
voxpelli
The documentation from
#
tantek
it is considered to be is a form of [[weasel wording]] that is sometimes used in specifications to justify features or assertions that have little or no basis in actual citable evidence or real world examples, because if it did then such citations would be cited!
#
voxpelli
That era seems to be lacking though
#
tantek
haha does Loqi not handle definitions of phrases with "is" in them?
#
Loqi
is done
#
tantek
apparently
#
tantek
gives Loqi a definition for is
#
Loqi
enjoys the definition for is
#
voxpelli
What one might want to do regarding the From-header is to continue the discussion on private data in PuSH
#
aaronpk
agreed, it seems like an extension or future development of PuSH and not related to the core at all
#
voxpelli
So that it eg. can work with public/private post upgrading https://indiewebcamp.com/private_posts#Public_Page_Upgrading
#
aaronpk
pull request filed to remove that phrase https://github.com/pubsubhubbub/PubSubHubbub/pull/41
#
aaronpk
voxpelli: re the signature, it's true the payload format allows a sort of extensibility by doing something like sha256=xxxx instead of sha1=xxxx
#
aaronpk
but the spec right now requires SHA1
#
aaronpk
"...where signature is a 40-byte, hexadecimal representation of a SHA1 signature"
#
voxpelli
aaronpk: yeah, just commented on the issue
#
aaronpk
that certainly would work. the only concern is that seems to go towards a path of not necessarily compatible implementations
#
voxpelli
not that many accepted algorithms out there? And one could still have a "must" on a fallback to sha1 if no better algorithms match
#
aaronpk
the problem with a MUST fallback like that is at some point sha1 will be trivially hackable and then it's like having no signature
#
aaronpk
but yeah there aren't toooo many accetable algorithms, and the whole industry tends to be aware of them at approximately the same pace
#
voxpelli
Other option is to do like OAuth 1 does and just allow anything
#
rhiaro
tantek, aaronpk: Between flights, I wrote about my post-type detection implementation: http://rhiaro.co.uk/2015/09/post-type
#
voxpelli
As long as there's a mechanism for algo-negotiation it should be possible for best practices to emerge I think
#
rhiaro
Aaaand now boarding the flight to Boston
#
voxpelli
OAuth 1 can't negotiate algo, hence almost everyone are stuck on sha1, including Twitter
#
ben_thatmustbeme
aaronpk: its an error case really, how do you handle u- on something that doesn't have a src= value
#
ben_thatmustbeme
unmung handles it equally confusingly
#
voxpelli
aaronpk: did you find any other issue with PuSH? One challenge with it right now is that it's a versions document rather than a living spec
#
ben_thatmustbeme
that was in regards to video tag thing from earlier
#
voxpelli
aaronpk: perhaps file an issue about making a living PuSH spec to take over after 0.4? So it can keep up with all other living specs
#
aaronpk
ben_thatmustbeme: oh you know what's happening with the php one? it's getting a blank value from the u- property, and then doing the relative URL resolution on it, resulting in the page's URL
#
ben_thatmustbeme
sort of tempted to do <span class="e-video"><video><source> ....
#
aaronpk
voxpelli: are we stuck on making progress on 0.4 because it's versioned?
#
ben_thatmustbeme
thats what i figured
#
voxpelli
aaronpk: versions specs carry the expectation of being stable so yeah, eg. HTML5 dropped the version number when it went living I think
#
tantek
voxpelli: not really. WHATWG dropped the "5". W3C embraced it.
#
tantek
so now we have both "HTML: Living Spec" and "W3C HTML5" (frozen feature set)
#
tantek
both are useful
#
tantek.com
edited /weasel_word (+26) "it is considered"
(view diff)
#
voxpelli
tantek: in relation to PuSH it feels like a living spec is what the IndieWeb community would want, right? But feed readers and the blogosphere probably doesn't want any changes
#
tantek.com
edited /it_is_considered_to_be (+34) ""it is considered" shorter version"
(view diff)
#
tantek
voxpelli: yes IndieWeb uses more of a "living spec" model since we're evolving / iterating things so rapidly
#
tantek
voxpelli: "doesn't want any changes" is a good way to become obsolete
#
aaronpk
even a lot of software development is moving away from the specific versioning
#
aaronpk
browsers are proud that they're able to push out updates to users automatically now, which means browser version numbers virtualy don't matter
#
voxpelli
aaronpk: you will probably receive less push back if the PuSH spec was made a living one, so suggesting it may make eg the algo change easier
KartikPrabhu joined the channel
#
tantek
aaronpk: except on caniuse.com :P
#
aaronpk
okay, do you think it's a reasonable suggestion for me to file an issue for that then?
#
tantek.com
edited /it_is_considered (+46) "citation for The man behind the curtain ;)"
(view diff)
#
tantek
sure - that will be interesting to see what happens
#
rhiaro.co.uk
edited /post-type-discovery (+162) "/* See Also */"
(view diff)
wolftune joined the channel
#
tantek.com
edited /it_is_considered (+74) "cite ietf.org search for it is considered to be"
(view diff)
#
tantek.com
edited /lulz (+23) "it is considered to be lulz"
(view diff)
Geng joined the channel
#
tantek.com
edited /lulz (+33) "see also"
(view diff)
#
aaronpk
so I went down this whole rabbit hole of figuring out a way to reliably store URLs mapped to a filesystem http://indiewebcamp.com/IndieArchive#Storage
emceeaich joined the channel
#
aaronpk
but now I'm thinking about dropping that whole idea and switching to something more like https://github.com/aaronpk/ca3db instead, which supports storing multiple versions of content easier
#
aaronpk
i'm not totally happy with either solution though
#
KartikPrabhu
aaronpk: thanks for documenting this. will be useful for my file storage plans
#
tantek
ooh I'm definitely interested in ideas around storing multiple version of content, in particular multiple versions of files and URLs
#
KartikPrabhu
aaronpk++
#
Loqi
aaronpk has 965 karma
#
aaronpk
i'm at the part of my site rebuild where I need to store comments and such
#
tantek
aaronpk: I have some notes on " storing multiple versions of files / URLs" here: http://indiewebcamp.com/wiki-projects#Brainstorming
#
KartikPrabhu
yes that was my use case as well
#
tantek
ironically, having personal wiki pages with saved versions is a bigger use-case for me than storing versions of comments!
#
vanderven.se martijn
edited /lulz (+15) "There is a Ben page on here too."
(view diff)
#
KartikPrabhu
aaronpk: any particular reason for adding ":" after domain name folder?
#
aaronpk
yes, i thought i documented that
#
aaronpk
could be clearer
#
KartikPrabhu
couldn't find reason "Append a : to the name of the folder (appears as / on OSX in Finder)"
#
aaronparecki.com
edited /IndieArchive (+258) "/* Storage */ make use of : more explicit"
(view diff)
#
aaronpk
the key is "Note that without appending ":" to folders, it would be impossible to store all these files on disk since "example.com/tags" would have to be both a file and a folder.", but basically chose ":" because it looks like / in Finder
#
KartikPrabhu
ahh gotcha
#
aaronpk
that's what broke in my current implementation, I started getting errors about not being able to create files then I realized what was happening
#
aaronpk
so hey it's not a theoretical problem :)
#
KartikPrabhu
is wondering if he can make everything a file instead of organising into folder heirarchy
#
KartikPrabhu
might not be best for IndieArchive though
#
aaronpk
you'll hit filesystem limits of max files in a folder tho
#
KartikPrabhu
oh crap! didn't think of that
#
aaronpk
not to mention "ls" or Finder hanging like crazy trying to list all the files
#
aaronpk
has some experience with too many files in a folder
#
KartikPrabhu
might want to add to that page then?
#
aaronpk
that's why mediawiki stores images in subfolders of their hash
#
aaronpk
if the hash is abcdefg it goes in a folder like a/b/abcdefg
#
tantek.com
edited /post-type-discovery (+125) "status, editor's draft, mature enough though for implementation and feedback!"
(view diff)
#
KartikPrabhu
man! file storage is hard...
#
aaronpk
computers are hard!
#
KartikPrabhu
lets use index cards
#
tantek
rhiaro++ for blogging about post type discovery!
#
Loqi
rhiaro has 165 karma
#
aaronpk
KartikPrabhu: hang on let me send you a reply on an index card via carrier pigeon
ttepasse joined the channel
#
KartikPrabhu
then i'll file it in my giant cabinet!
j12t joined the channel
#
tantek.com
edited /post-type-discovery (+20) "link implementations"
(view diff)
j12t_ joined the channel
#
aaronpk
hm wondering if it's possible to turn my post type discovery code into a library
#
aaronpk
the trick is what serialization or object would it work against? parsed mf2 json I guess?
benwerd joined the channel
#
fiatjaf.alhur.es
edited /Webmention (+267) "add jekmentions as a service for receiving webmentions"
(view diff)
#
Jeena
I wonder if there is something in microformats to describe my relation to my band and to friends with their own domains (blogroll!)
emceeaich joined the channel
#
tantek
Jeena - to friends in your blogroll - that's precisely the use-case that XFN ( gmpg.org/xfn ) was designed for!
#
Kongaloosh
tell tbrb where are we meeting this week?
#
Jeena
interesting that I never heard of XFN before
#
KartikPrabhu
aaronpk: just received your index card via pigeon! thanks
#
tantek.com
edited /plurality (+86) "emojicon"
(view diff)
#
Jeena
so if I link to the IndieWebCamp website I theoretically could also put a rel="member" on that link?
#
Jeena
oh no, rel="group" I mean
#
tantek.com
edited /plurality (+216) "expand dfn"
(view diff)
#
tantek
true!
emceeaich, squeakytoy2 and indie-visitor joined the channel
#
Loqi
Welcome, indie-visitor! Set your nickname by typing /nick yourname
#
indie-visitor
akolonin
#
tantek.com
edited /Diaspora (+178) "/* Criticism */ monoculture evidence, Diaspora homepage links to a singular codebase repo"
(view diff)
#
kylewm
what is a living spec?
#
Loqi
It looks like we don't have a page for "living spec" yet. Would you like to create it? http://indiewebcamp.com/s/103k
#
kylewm
(genuine question)
#
tantek
kylewm: you may want to ask in #whatwg ;)
#
bear
remember that even index cards are usually stored in a partioned manner
parzzix joined the channel
#
tantek
bear, as are punchcards ;)
#
bear
gets all sentimental remembering his punchcard "library" of routines
#
kylewm
Looks like living spec means it is updated continuously based on feedback from implementers and consumers
#
KevinMarks
why is there no random() in CSS?
KartikPrabhu joined the channel
#
aaronpk
<3 bridgy facebook notification that authorization is expired
#
aaronpk
sadface that I don't check my facebook notifications very often :P
friedcell, emceeaich, originalbf and [snarfed] joined the channel
#
[snarfed]
aaronpk: aww thanks. subscribe to your fb notif rss feed!
#
[snarfed]
i love that they have that. they'll probably kill it any day now, but still
#
aaronpk
oh ha good point
#
aaronpk
I will once I finish my (new) reader
#
[snarfed]
i guess i could also email when it expires... but... no
#
aaronpk
no no not necessary lol
cmhobbs joined the channel
#
KevinMarks
tantek, you'll know this: why is there no random() or access to current time in CSS? Cam only do deterministic animations because of this
#
tantek
I have a feeling, uh, because crypto? ;)
#
aaronpk
cause computers can't be random :P
#
aaronpk
except when you don't want them to be
#
tantek
someone tries animating a virtual lava lamp, someone reads somewhere that you can do crypto by sampling a lava lamp. hilarity ensues.
#
@kevinmarks
This is a great use of SVG to make a lava lamp http://codepen.io/chrisgannon/blog/how-to-make-an-svg-lava-lamp but could you do it with SVG animation, not js? @tabatkins?
(twitter.com/_/status/649270771148173313)
#
KevinMarks
so even though css+html is turing complete, we can't do a PNRG because we don't have a way to seed it?
shiflett joined the channel
#
KevinMarks
or can't make a clock in pure CSS, have to use JS to look up the time.
#
aaronpk
know what blew my mind the other day? the calc() function in css.
#
KevinMarks
can you get a random number out of CSS by using calc() to count things that may not have loaded yet?
#
aaronpk
ever since I started learning css I always wanted that to exist
#
aaronpk
and apparently it's been there for a while, so I guess I need to go back to school or somethign
#
ShaneHudson_
Quite a few things like that have been there for a while but with poor support. It is all getting better now :)
#
ShaneHudson_
I was amazed yesterday that everything from IE8+ supports counter
originalbf joined the channel
#
aaronpk
i don't even know what that is
#
ShaneHudson_
Basically a way to access the counter of generated content (ie. a list). So you can reset the numbers etc. Basically one of the first CSS variables.
#
ShaneHudson_
I used it to hide the default numbers and add them back in as a psuedo, so that it could be styled further than usual
#
ShaneHudson_
Also, current-color is a really nice one that has pretty good browser support now, lets you get the current text colour. So you can create nice underlines (with a gap for hanging letters) using shadows etc with the colour set to the current. Useful, though hacky
#
KartikPrabhu
I use current Color for SVG icons
benwerd, lukebrooker, Geng, KartikPrabhu, elf-pavlik, frzn and yakker joined the channel
tantek, friedcell, gavinc and emceeaich joined the channel
emceeaich and shiflett joined the channel
#
@artwisanggeni
#python went 0.1.3: Tools for implementing a webmention enpoint. https://pypi.python.org/pypi/went/0.1.3
(twitter.com/_/status/649358599412580353)
wolftune and KartikPrabhu joined the channel
#
KevinMarks
where is the micropub posting html vs plaintext written up?
tantek joined the channel
#
KevinMarks
nm, found it
KartikPrabhu and lukebrooker joined the channel
#
acegiak
kevinmarks, lol and like, every pre-2007 website
#
KevinMarks
yours is nice and subtle though
#
acegiak
there's a scene atthe end of serenity where the graves have looping bits of video holographically projected and it's subtle but lively and nice
#
acegiak
which is what I want gif profiles to feel like
#
acegiak
or at least mine :