#indiewebcamp 2011-07-17

2011-07-17 UTC
aaronpk and neophiliac joined the channel
#
brennannovak
heya neophiliac: I'm just pushing up a readme file regarding my idea Later.js
#
neophiliac
brennannovak: c'mon, man, I'm aging! (what movie is that from?)
#
brennannovak
hmm... maybe "come on man, I'm f***in dying over here" from Resevoir Dogs?
#
brennannovak
Ok here is my rough repo for Later.js https://github.com/brennannovak/Later.js
#
neophiliac
brennannovak: Nope, Americathon with John Ritter as POTUS.
#
brennannovak
hehe
#
neophiliac
brennannovak: It's VERY old, and was bad (in a good way) when it was new. It might not be so entertaining now.
#
neophiliac
Or it might be better, who knows?
#
brennannovak
kinda like Airplane?
#
neophiliac
brennannovak: I like the idea, but I'm still not getting my head fully wrapped around the use cases.
#
neophiliac
brennannovak: Wait, what if the client pulled your connections (or whatever) from all CDNs, and made the connections (and drew the graph) locally?
#
neophiliac
No, it was much worse than Airplane.
#
brennannovak
so in my current implementation of SocialIgniter when i post content my server install will syndicate content to other networks (some with proprietary API's like Facebook, Twitter...) and others installs of the same app (Social Igniter)
#
brennannovak
but the problem is the amount of time it will take to syndicate to other networks as the number of N networks scales to high levels
#
brennannovak
also availability of said network
#
brennannovak
status updates I try to post to Twitter are constantly not properly syndicating cause their api is down
#
singpolyma
brennannovak: the real solution is NNTP (or something similar)
#
singpolyma
instead of this broadcast-to-everyone model
#
brennannovak
NNTP ?
#
neophiliac
brennannovak: OK. I can see that. So e.g. I publish a blog post and the response is some JS that tells my browser to post notifications to twitter, FB, G+. etc?
#
singpolyma
The Usenet protocol. It was built for massive federation over unreliable nodes
#
neophiliac
singpolyma: yeah, NNTP is an old P2P protocol, right?
#
singpolyma
well, he seems to be suggesting farming that out to users' browsers as well
#
neophiliac
Haven't read the RFCs for a lot of years...
tantek joined the channel
#
singpolyma
neophiliac: yes. just after UUCP ;)
#
brennannovak
yes, farming it out to the browser
#
singpolyma
but it still solve the blog/microblog protocol problem very well, beacuse that's exactly what Usenet is/was: lots of little pubsub groups
#
singpolyma
brennannovak: if you're farming out anything that needs auths (like syndicating to Twitter) then you're exposing your credentials to random users
#
neophiliac
Moving the intelligence to the edges...
#
singpolyma
Also, the processing/bandwidth it takes to get this to the user seems like it would be the same or more than just doing it yourself, no?
#
neophiliac
singpolyma: except that the user's browser is already authenticated, so the notifications it sends will include the locally-stored auth.
#
brennannovak
yes precisely, the auth tokens would only be the user in question who's content is being syndicated
#
neophiliac
singpolyma: that is, the web app doesn't have to send that, only the list of places to notify
#
singpolyma
neophiliac: if we're talking about the posting user's browser federating only his posts, but that doesn't seem likely to be useful at all. since that's just one machine
#
singpolyma
Hmm... maybe. I'm not sure the user's DSL connection on his netbook is more reliable that the real server on a good internet line that the requests starts on ;)
#
brennannovak
no it's just a matter of reducing load on the server
#
neophiliac
singpolyma: if the user can click the 'publish' button and get a response, then the internet connection is probably good enough. :)
#
singpolyma
It seems like you're preoptomizing. Is there any evidence that this sort of microoptomization will matter on any real-world servers?
#
neophiliac
brennannovak: so this is going to benefit servers that have a lot of outgoing connections as a result of a user's request. And a lot of traffic.
#
brennannovak
if I setup a social site and it starts to get a fair amount of concurrent users posting content, the syndication of that content could become quite hairy if the follower count of those users (on said network) increases. The goal is to offset some of that work to each users browsers
#
neophiliac
singpolyma: I agree, but it's an interesting exercise, and could be fun to test.
#
singpolyma
oh, sure. I'm all for building things just because they're interesting :)
#
brennannovak
hehe
#
neophiliac
brennannovak: Now you need a cool acronym, preferably a cleaning compound. I suggest making MRCLEAN fit, because the logo would be so awesome.
#
brennannovak
on my personal node of Social-igniter my messages to Twitter sometimes don't go through, the goal is build a job que so that those messages will get syndicated at a later date when the API is up
#
brennannovak
neophilac: how about Borax
#
brennannovak
;)
#
neophiliac
Isn't borax an ingredient in flubber?
#
brennannovak
yes
#
neophiliac
brennannovak: you'd need a way to persist the job list, because the user might shut down or sleep the laptop before all messages are sent. JS doesn't make that easy.
#
brennannovak
true
#
singpolyma
offline storage
#
singpolyma
but that's not reliable
#
brennannovak
I was planning on using Local Storage
#
singpolyma
yeah. but then you're relying on them using the same browser next time they go to the site where the JS is
#
singpolyma
which may be fine for you use case
#
tantek
should your server handle the store and forward rather than your client when the downstream API is down
#
brennannovak
I rarely switch browsers and visit the same site
#
singpolyma
tantek: normally. normally the server should handle everything :)
#
singpolyma
brennannovak: interesting. I often do, but that's because I use 4 devices
#
brennannovak
signpolyma: I wasn't taking into account multiple devices
#
brennannovak
sooo... Local Storage for the job que (and handling of jobs) and a backup of that que stored in users profile data if the user opens the site on more than one device, but that's getting a bit hairy
#
aaronpk
I'm coming in to this conversation late, but what is the benefit of doing it in a browser?
#
singpolyma
aaronpk: for fun
#
brennannovak
aaronpk: to offset server load & handle failure to syndicate occurrences
#
singpolyma
I'm still unconvinced of the "offset server load" argument. It's a bit like a MacTruck saying to a Prius "here, take some of my stuff because I have a lot of it"
#
aaronpk
seems like handling failures can be done better on the server side
#
neophiliac
singpolyma: more like the truck saying that to 1000 priuses...
#
brennannovak
also, for environments and user cases where users have no idea how to setup cron jobs
#
neophiliac
If you have one client per server, you have bigger problems...
#
aaronpk
wordpress has done a pretty good job of faking cron tasks
#
singpolyma
neophiliac: I'm confused. I thought the idea was for just the posting user's browser to do this? and that's why it wasn't a security issue ...?
#
singpolyma
brennannovak: you can always use wordpress-style fake cron
#
brennannovak
how do they do fake cron jobs?
#
brennannovak
s/do they/does wordpress
#
singpolyma
Every time a user hits the page they open a socket to the wp-cron.php script, and it decides if stuff needs to be run right now and does so (server side)
#
neophiliac
singpolyma: assume that the cargo on the truck belongs to the Prius drivers anyway. :)
#
singpolyma
neophiliac: Sure. But if I have a truck, a Prius, and some cargo, I don't say to myself "hmm, I wonder how much I can stuff into the Prius... don't wanna overload the poor truck"
#
neophiliac
singpolyma: the truck is just giving some of it back when the hill gets steep. :)
#
brennannovak
yes, it's a matter of the size of the load
#
neophiliac
singpolyma: OK, this might be ridiculous, but the truck has a fixed capacity. As it accepts more loads from Prius drivers, it sends a little bit of the load back so that it doesn't exceed its capacity.
#
brennannovak
small networks, with a few users with a few followers, no biggy, the truck can handle
#
brennannovak
I love neophiliac analogies, I think he gets where I'm going with this :)
#
singpolyma
big networks with many users with many followers... get bigger servers. I mean, I'm all for swarm, and for trying stuff, I'm just unconvinced that delegating tasks for one user to that user's browser will get you much, if anything
#
brennannovak
I wanna solve the issue of my messages syndicating to twitter even when their api craps out... and for some reason I wanna attempt to do it in local storage and JS rather mysql and php and cron job
#
neophiliac
brennannovak: I didn't come up with this analogy, I'm just running with it. :)
#
neophiliac
singpolyma: my servers are on a 100M drop, 1 hop from a 10G connection to PAIX and MAE-WEST.
#
singpolyma
neophiliac: sounds pretty good
#
neophiliac
singpolyma: not many broadband connections can handle as much load. But this is interesting, and offloading some of the joins could be a fun test.
#
brennannovak
aaronpk: yah, that wp-cron is a nifty solution and sort of how I was seeing this idea working
#
neophiliac
I'm thinking of FB's eventual consistency... some of the joins could be done in the browser. Maybe a multi-browser vote?
#
brennannovak
interesting
#
neophiliac
singpolyma: With FB's load, this could reduce their server load measurably. Lots of risk, but hey, it's an experiment.
#
singpolyma
neophiliac: Like I said, not convinced, but I'm willing to be wrong :)
#
neophiliac
singpolyma: I'm not saying I'm convinced of anything more than it would be fun to build and play with. :)
#
singpolyma
for sure :)
#
brennannovak
coolio, thanks y'all :)
#
brennannovak
I'll start tinkering with it soonish at https://github.com/brennannovak/Later.js
#
neophiliac
brennannovak: cool. I'll watch it.
dbounds1, Loqi, dbounds, brennannovak, gazoombo, abki_ and abki__ joined the channel
#
Loqi
[http://twitter.com/gazoombo] .@t Do you know of any #indieweb servers that consume data from last.fm clients? I wish I had a pass-through. cc: @mxcl @singpolyma
#
gazoombo
hey, that's a cool trick. I was about the cross-post here too.
brennannovak_, devinrolsen and quartzjer joined the channel
#
tantek
hey gazoombo - I don't - though the last.fm use-case is an interesting one to consider from an indieweb perspective
#
tantek
i.e. how would you "scrobble" directly to your own website, and then syndicate your scrobbles to last.fm
wchulseiee joined the channel
#
wchulseiee
hi all
abki_ joined the channel
#
gazoombo
tantek: exactly. scrobbles could also be syndicated out as Activity Streams and you could even filter it first. For example you could run each item through a check 'if x not in guilty_pleasures' or what have you
#
tantek
hahaha - totally an excellent use-case for indieweb scrobbling (above last.fm or any other social silo scrobbling)
#
tantek
gazoombo - could you copy your brief thoughts (like even just a sentence or too) on that to http://indiewebcamp.com/scrobbling
#
gazoombo
because maybe I want to someday generate a chart of how often I indulge in guilty pleasures, but I don't want the world to know what they are
#
gazoombo
certainly
#
99.180.73.73
edited /Special:Log/newusers () "created new account User:Gazoombo"
(view diff)
ewaldl joined the channel
#
gazoombo
created /scrobbling (+595) "Introduce the concept of 'IndieWeb Scrobbling'"
(view diff)
#
gazoombo
edited /scrobbling (+403) "Add ideas for advanced features: alt. syndication & filtering"
(view diff)
#
gazoombo
edited /scrobbling (+41) "add link to definition of scrobbling"
(view diff)
#
gazoombo
edited /scrobbling (+42) "add link to Last.fm API doc"
(view diff)
gazoombo, devinrolsen, quartzjer, krijnh and brennannovak joined the channel