#dev 2017-11-10

2017-11-10 UTC
eli_oat and snarfed joined the channel
#
oodani
Found the bug breaking Omnibear - grantcodes/micropub doesn't work properly if you have a <base>, like I do. I'll make a pull req. :)
tantek joined the channel
#
grantcodes
oodani: I think it should be fixed with the latest version of the micropub-helper library
snarfed1 joined the channel
#
oodani
I'm not so sure, I checked the relevant part of the master branch and it still seemed to have the same issue. I've made the pull request, take a look?
#
grantcodes
There was a fair bit of discussion about base tags
#
grantcodes
Ok will do thanks ?
#
oodani
It was basically just Incorrect JavaScript™ rather than necessarily handling base tags in a semantically wrong way. ¯\_(ツ)_/¯
[kevinmarks] joined the channel
#
grantcodes
Ah cool, I'll merge that soon! Base tag support is very new in the library!
#
aaronpk
wait does Quill follow the <base> tag properly? I don't remember writing any code for that
#
oodani
I'm not sure, since not doing anything with the <base> tag is equivalent to correctly supporting it in my case. :P
#
oodani
Quill works very nicely, but it could just be ignoring <base> entirely and would still work fine.
eli_oat, dougbeal|mb1, tantek, renem, [tantek] and [chrisaldrich] joined the channel
#
eddiehinkle.com
edited /Indigenous (+261) "/* IndieWeb Interest */ +beta info"
(view diff)
tantek, j12t, KartikPrabhu, cweiske and barpthewire joined the channel
#
www.boffosocko.com
created /personas (+624) "prompted by [chrisaldrich]; basic definition; links to IWC sessions"
(view diff)
#
Zegnat
“not doing anything with the <base> tag is equivalent to correctly supporting it in” - ha, that’s an interesting test case!
#
Zegnat
oodani 7.5 hours ago. I did a lot of BASE element stuff last weekend but never thought to test with a BASE element that didn’t have any effect.
#
cweiske
you're talking about <base href=".."/>?
#
Zegnat
Yes, that’s what the discussion last night seemed to have been about.
#
Zegnat
I just continued on the last messages in this channel, but didn’t realise the last messages were over 7 hours ago
#
www.boffosocko.com
edited /Indieweb_for_Education (+345) "Aaron Davis article "My IndieWeb Reflections""
(view diff)
#
www.boffosocko.com
edited /generations (+3141) "links in see also; list of related IWC sessions with brainstorming; additional brainstorming thoughts under gen2; section for generation examples and notes"
(view diff)
KartikPrabhu, loicm, barpthewire and [kevinmarks] joined the channel
#
@Lady_Ada_King
Tempted to do a talk of 'stuff I like' with comlink, indieauth, aframe, web midi with an underlying theme of declar… https://twitter.com/i/web/status/928987121339240448
(twitter.com/_/status/928987121339240448)
tantek and snarfed joined the channel
#
aaronpk
cweiske: hmm I thought that superfeedr should be subscribing to your hub and then it'd show up here via Loqi. Tho I'm not quite sure how to tell superfeedr about your blog in the first place.
eli_oat joined the channel
#
oodani
cweiske, the <base> I use on 00dani.me is actually <base href="https://00dani.me" />, which is precisely what browsers would be using anyway. (At least for URLs relative to the site root, but all my relative URLs are in that form so. Yeah.)
#
oodani
The only reason it's there is so that saving HTML from my site, then parsing it later, still resolves relative URLs correctly.
#
oodani
For example, I run commands in the form "https 00dani.dev | mf2" fairly often during development, which is useful because I've taught HTTPie about my self-signed dev certificates and so it all works out. :3
j12t joined the channel
#
aaronpk
wow I really messed something up when I repost stuff that has html
#
aaronpk
weird, it must have something to do with a recent change cause all my old reposts are broken too
eli_oat joined the channel
#
aaronpk
fixed. that was sneaky
#
aaronpk
code reuse is good, until you forget all the different places the code is used
eli_oat joined the channel
#
loqi.me
created /BBS (+172) "prompted by petermolnar and dfn added by petermolnar"
(view diff)
#
kaja.sknebel.net
edited /BBS (+1) "linkify ('x is y. <url>.' pattern)"
(view diff)
#
dgold
aaronpk: curious -- what should a micropub server return in the case of a Nested Microformat Object?
#
aaronpk
return for what request?
#
dgold
if the server _accepts_ the Object, but the Object as saved won't render in a deliverable fashion
#
dgold
then what can the server return? 204?
#
aaronpk
hold on back up
#
dgold
(micropub.rocks test 204 - amusingly)
#
aaronpk
you want to reject the request?
#
dgold
No, i'm trying to work on it to get my endpoint to accept it.
#
aaronpk
also it would seem this question would apply to more than just nested microformats objects, for example if the client sent a property you didn't recognize and couldn't do anything with
#
aaronpk
if the endpoint is creating a post, then the response should be "201 created" and include the URL of the post
#
aaronpk
that's the simple answer
#
dgold
right, but this isn't creating a post - its creating a non-post Object, example given is a h-measure.
#
aaronpk
there's still a post around that h-measure
#
aaronpk
the post has a summary
#
dgold
huh. that's a value that's not included anywhere else in the test suite
#
aaronpk
hm true
#
aaronpk
you know what would be a better example here is probably the checkin object that ownyourswarm sends
#
aaronpk
since that looks like a normal post and also includes a "checkin" object
#
dgold
i've just been simplifying nanopub a _lot_, and thought I could usefully try and grab that dangling red-mark
#
dgold
yes, you're right
#
aaronpk
I used h-measure in the test suite because I hadn't figured out checkins on my site yet, and h-measure seemed like a simpler example.
#
aaronpk
although at this point there are now quite a few implementations that support checkins, and almost none that support h-measure
#
dgold
oh, by the bye, and apropos not very much: I'm using the media endpoint implementation you were kind enough to share
#
aaronpk
so it would seem appropriate to use checkins as the example for the test suite
#
aaronpk
oh cool. did I post that somewhere? I forgot
#
dgold
and for some damnable reason, it refuses to accept .gif files
#
dgold
.png, .jpg, all the rest - perfect. .gif, nope
#
dgold
Yes, and I just realised I don't have the source url.
#
dgold
I'd agree that a check-in would be a better representative example. I've been making nanopub start storing the lat long details now.
#
aaronpk
hmm it seems to have posted the gif just fine on my local copy of the media endpoint
#
aaronpk
I should probably add that media endpoint source to actual version control
snarfed joined the channel
#
aaronpk
anyway what's happening when you're posting a gif?
#
dgold
on the server side, it says Unrecognized. It returns a location, but doesn't save the gof
#
jackjamieson.net
created /User:Jackjamieson.net (+517) "Created page with "<span class="h-card">[http://jackjamieson.net/ <img src="http://0.gravatar.com/avatar/c850ec9240a04bd1ebf02540e4f57b73?s=125&d=default&r=g" style="width:125;float:right;margin-le...""
(view diff)
#
aaronpk
"Unrecognized"? and this is my media endpoint code?
#
dgold
"PHP message: PHP Warning: copy(): Filename cannot be empty in /opt/media/micropub.php
jackjamieson joined the channel
#
dgold
oh, no, that was a failure handler I wrote
#
dgold
its your failure handler
#
dgold
fwrite($fp, "Unrecognized: $type\n");
#
aaronpk
oh gotcha
jackjamieson joined the channel
#
aaronpk
what did it write to the file? it's supposed to write the mime type it didn't recognize
#
dgold
nothing - Unrecognized:\n
#
aaronpk
hm k. it's supposed to use the file extension given after that, and fall back to .txt
#
aaronpk
I don't see how $filename could be blank
#
dgold
maybe its a failure on micropub.rocks?
#
aaronpk
oh maybe the copy() error is saying the $file['tmp_name'] is blank
#
aaronpk
can you print out $_FILES and see what's there?
#
aaronpk
ohhh you know what
#
aaronpk
check your max upload filesize in php.ini
#
aaronpk
the gif is >2mb so if your max upload size is 2mb it'll fail
#
dgold
we have a winner!
#
aaronpk
and I just replaced test 204 with a checkin
#
dgold
hoorah!
#
dgold
I think that's me with everything except the multiparts
#
dgold
sea of green checkboxes!
snarfed and eli_oat joined the channel
#
aaronpk
[eddie] any chance of getting some instructions for how to build Indigenous from source? It looks like it's using some dependency management stuff I'm not familiar with.
KartikPrabhu and snarfed joined the channel
#
snarfed
"Twitter's API is the internet's worst joke, Instagram seems to have an API mainly to show what outside developers can't do, and Flickr is a hollowed-out shell of its former self." ???
eli_oat joined the channel
#
sknebel
in that context the mention of Flickr seems kind of out of place
#
sknebel
I thought Flickr was just left alone mostly the last years, without much innovation or breakage?
#
snarfed
or maintenance
#
petermolnar
today's CSS lesson: opacity settings make things funky when printing to PDF
#
aaronpk
no they keep changing the web UI
#
aaronpk
and it's gotten worse and worse each time
#
sknebel
oh, ok
#
dgold
aaronpk: how did you get the stacktrace for the iOS crash in Indigenuous?
#
aaronpk
opened up xcode with my phone plugged in and watched the logs
tantek joined the channel
#
aaronparecki.com
deleted /testtesttest "content was: "{{stub}} '''<dfn>[h testtesttest]</dfn>''' is something.""
#
sknebel
whops, did I forget to clean that up? sorry :/
John___ joined the channel
#
schmarty
[eddie]: getting a crash when trying to sign in on just-installed indigenous. My site uses IndieAuth for auth. I tried email code, Twitter, and Github, same results.
#
dgold
has just realised the behavious I saw this morning is a crash, and not a persistence error
[kevinmarks], wagle, snarfed and [eddie] joined the channel
#
[eddie]
aaronpk schmarty dgold: Yeah apparently something went wrong with the build. It's been working on my phone fine, but when I installed it from TestFlight it definitely is having the same issue as all of you
#
dgold
could be somthing apple did?
#
[eddie]
When it goes to TestFlight, it compiles to bitcode and then that gets built on the fly for the specific device you're downloading to, so some of my code might not be working right with that
#
[eddie]
!tell aaronpk: I haven't pushed the latest code up to github. I'll try to get that pushed up. The dependency library I was using wasn't working right so I decided to shift to a simpler method of just copy and paste the dependency as a folder for now. So I'll try to get that new method up onto GitHub
#
Loqi
Ok, I'll tell them that when I see them next
#
[eddie]
Eventually the plan is to use carthage (that dependency library that the github code was using) but that would error out when uploading to iTunes connect but install fine on my local device
#
[eddie]
It's so strange how different the app can be between deploying to a test machine and uploading through iTunes ?
#
[eddie]
I guess it's kind of like a dev machine and prod machine in web dev ;)
#
aaronpk
my app is very simple and I didn't have any issues like that
#
aaronpk
it's not Swift, and doesn't have any fancy dependency management stuff tho
#
[eddie]
Yeah it is very strange. I've been gone all day but I'm looking forward to sitting down and taking a look at my logs this evening and seeing what's going on with it
[jeremycherfas], eli_oat, snarfed, tantek, [miklb] and [kevinmarks] joined the channel
#
aaronpk
I'm getting pretty overwhelmed with the idea of consuming h-feed again
#
aaronpk
There seem to be two main reasons
#
aaronpk
1) turns out there's a huge variety of structures of h-feed once you go and look at even a handful of sites. Compare Tantek's home page to mine to a Known site
#
aaronpk
2) you can't rely on just the html of the h-feed, since some people's lists of posts show only partial data and you have to fetch the post permalinks to get the full post data
#
sknebel
what is h-feed?
#
Loqi
h-feed is a microformats2 experiment with a top level feed object to contain h-entry posts https://indieweb.org/h-feed
#
aaronpk
I can smooth over #1 with a bunch of work and write it into XRay
#
aaronpk
but #2 is a harder problem
#
snarfed
2 is harder? it seems like a lot more compute and network load, but much less dev work than 1
#
www.svenknebel.de
edited /User:Www.svenknebel.de (+16) "/* short-term itches */ need feeds"
(view diff)
#
snarfed
also fwiw bridgy also has to handle many of these problems for its original post discovery
#
aaronpk
It also doesn't play well with WebSub
#
snarfed
true...but you can't really require websub in a reader anyway, so if you already have to implement polling...
#
snarfed
one piece of prior art for finding/normalizing h-feed: https://github.com/snarfed/mf2util/blob/master/mf2util.py#L748-L787
#
aaronpk
I don't have to implement polling in the reader if there is a shim that implements polling for me and turns it into notifications
#
aaronpk
e.g. superfeedr
#
snarfed
ah yes true!
#
snarfed
superfeedr++
#
Loqi
superfeedr has 1 karma in this channel (9 overall)
#
aaronpk
but then I can't necessarily use the contents of the WebSub notification cause I'll have to go fetch each post permalink too
#
snarfed
yeah but it at least removes the polling, which is arguably the bigger dev + ops task
#
snarfed
also it's definitely nice that superfeedr + microsub (hopefully) dramatically reduces your work to build a reader server
#
aaronpk
i was trying to sit down and start the Microsub server by letting superfeedr handle all the polling for me, but then started getting overwhelmed by all the h-feed variations
#
snarfed
try just blindly translating one of those two examples, see how far it gets you
#
aaronpk
puts on his python glasses
#
aaronpk
Does woodwind fetch post permalinks or just use the feed?
#
snarfed
¯\_(ツ)_/¯
leg joined the channel
#
aaronpk
snarfed: I don't see how that code handles Tantek's case where he has an h-card -> h-feed -> items
#
snarfed
aaronpk: mf2util's? or bridgy's?
#
aaronpk
Either
#
snarfed
bridgy uses find_all_entries for h-feeds, which searches the entire tree of all child items
#
aaronpk
ah so it recurses?
#
snarfed
mf2util use find_first_entry, which uses find_all_entries (which recurses) under the hood