#aaronpkok changed the query on the IRC logs to use one of the indexes, so the pages should be a lot faster and should also not crash the server when a crawler goes nuts
#tantekFrom KevinMarks 's URL: "going to get backlash for the Google Analytics code I threw in there (when I suspected it was going to get attention). It shows that the philosophy it preaches is impossible to actually implement in the real world" - bullshit. Never had google analytics slow JS includes on my site - have yet to be convinced of any ROI of including it.
#KevinMarkseasier than running your own log analysis
#tantekI'd rather start with a stats package that gave me decent analysis of my apache logs - and I don't even have that. (OSS apache log analyzer suggestions welcome - heck we should have indiewebcamp.com/analytics )
#brianloveswordsSits at the front of a webserver and routes requests to stuff on the internal network (in this case, sockets or ports running on the machine)
snarfed and bnvk joined the channel
#tantekwow ads on youtube, even for small independent vids, is getting quite annoying
#tantekalso - they stopped putting link rel shortcuts on their permalinks :(
#tantekwell I have completed design sketches of what I want event permalinks and event summaries in lists to look like, so now I'm working on the data storage for the information needed for those designs
#bretI keep the data in yaml headers as part of a note, and use a form to enter the event data. It would be rad to do natural language event creation eventually
#tantekI'm just putting the event info directly into the content as an h-event
caseorganic, andreypopp, melvster and b0bg0d joined the channel
#bearwhen looking for webmentions in my post, does the spec allow for rel="webmention" or do I need to follow every external link to discover if it's a post?
#bearis writing a python webmention tool for sending and receiving
caseorganic, LauraJ, Jihaisse and friedcell joined the channel
acegiak, b0bg0d, DanielHeath, Nadreck, abrereton and julian` joined the channel
#julian`anyone know how i can get the text on my site to be vertically centered? atm im using <br>s but it doesnt seem to look good in firefox only chrome
#aaronpkI feel like some of the internals of p3k are getting a little messy, and stricter templating may help clean it up
#barnabywaltersI started implementing a DOM templating library, then realised I was just recreating jQuery in PHP, then decided other things had higher priorities, e.g. getting the shiny new taproot live
#neuro`Templating is always a mess. We had a loooong argument about that on Publify back in 2006 that eventually led to an ephemeral split of the projcet
#aaronpkby templating I mean not inventing a new templating language
#neuro`Some people wanted to use Haml, some other ERB.
#bearaaronpk - in the body of a post, I was curious how to know that a <a href…/> pointed to an external post that needed to be processed as a webmention. I realize now all of the PHP code samples i've seen use rel="in-reply-to" for that
#barnabywaltersbear: publishers should send webmentions to all URLs a post replies to (u-in-reply-to or rel-in-reply-to) and all URLs mentioned by a post, e.g. in the content
#aaronpkbear: still not sure which direction you're talking about
#barnabywaltersignoring URLs of assets like images is probably fine
#bearI have written a post and I'm running my webmention-o-rama code against it, how do I know what links to consider
#bearah, so I should be scanning all external references (sans assets0
#aaronpkthink of it this way, your post contains links to a bunch of external sites. if you publish your post, they can eventually find your post. the webmention just helps that process happen faster.
#barnabywaltersweb*mention* is named literally — it’s to let a URL know that it’s been mentioned somewhere
#beark, I just figured webmentions was more of a asserted thing instead of a passive thing
#waterpigs.co.ukedited /webmention (+275) "/* Which links should receive webmentions */ added basic answer to which links question, added should assets be webmentioned question, implementation documentation" (view diff)
#barnabywaltersaaronpk bear: ^^^ I added to the FAQ, feel free to refine/add issues/more implementation documentation
#snarfednot sure i currently translate those to AS and MF2 correctly, but i'm putting it on my todo list right now
#snarfedtwitter too. under the covers, it's a link, but UX wise, i expect more users think of the images as "embedded" than as "linked"
#snarfedre what the receiver would do, as a first pass, it's probably reasonable just to embed them as remote images, ie with the src pointing to the source's image
#snarfedmore sophisticated receivers could download the image and point to the local copy
#tanteksnarfed, in the FB scenario - the image is from the link
#tantekso if you're already sending a webmention for the link ...
#snarfedwere you asking, when people write a reply that they plan to send a webmention for, should they ever include embedded images (or video etc) in that reply?
#tantekwebmention is typically used today to communicate: " hey your (target URL) path was linked to from here (source URL)"
#tantekso I think this is one of the reasons Apple is switching to only iCloud-based syncing
#tantektheir syncing code is so bad that it can't handle changes in two devices
#tantekby syncing with iCloud, they treat iCloud as "truth" and therefore assume that even when you change things on one device, before you can change things on a second device, the first device syncs its changes to iCloud
#tantekand then the second device gets those changes before you can make changes there.
#snarfedat least when you get enough of it right that you only prompt rarely enoughy
#tantekbear, snarfed - I call bullshit. we've had multi-master merging working in many cases automatically in MediaWiki, github, mercurial etc. for *years*
#snarfedagreed that multi-master, paxos etc are different than ostatus, etc in that the algorithms themselves aren't necessarily complicated, it's just that implementing them is, esp at scale
#snarfed(paxos is definitely baroque, but not quite the same thing as complicated)
#tantekbear, yes, per-type sync makes sense because user expectations are different per-type.
#tantekso it looks like iTunes does sync by treating the Mac as master, so if you delete something on your Mac, but add to it it on your iPod/iPhone - it gets deleted next time you sync, and everything you added is thus lost.
#XgFtantek: And, in those multi-master merges that work, you need to hold the entire history (or at least enough), and still it needs user help depressingly often
#tantekI guess what I don't understand is the design decision to assume a delete even when there's a conflict. I'd think a fork would be better.
#tantekit's easy to delete something one more time. harder to undelete.
#tantekXgF, true about holding "enough history", though even just 1-2 revs back is enough to do plenty well (across *personal* devices, not group workflow)
#tantekI'm more talking about how to handle the error/collision cases
#tanteksadly, it looks like iTunes didn't keep the previous backup - that had my notes, and I can't see anyway to tell iTunes - please keep 2-3 backups of this iPod.
#XgFRouting everything through a central authority is a lot easier than retrofitting versioning to all those apps, also in most cases what you get are that things are instantly consistent (push notifications)
#tantekso no, all that automatic network traffic they do sucks down the battery so everyone I know with a decent Android has to recharge them late at night. Everyone.
#XgFIs recharging your phone overnight unreasonable?
fmarier joined the channel
#XgFtantek: I can tell you that Calendar/Contacts/E-Mail sync is completely negligible in battery impact on my phone...
#tantekThat sounds like something that requires a post to substantiate - lacking that, I don't believe you (my experience with network syncing/battery use is in direct contradiction, across iOS, Android, MacOS)
#XgFtantek: Android has an exhaustive battery usage monitor. Sitting right at the top is display. Sitting pretty far down are Data and Google Play Services
#tantekhowever, running out of power and requiring an external battery (what I see Android users do) is real
tantek joined the channel
#tantekSuch mobile displays (battery, signal strength, 2/3/4G) are "feel good" displays that manufacturers tweak/fake to make devices seem better than they are. iPhones were notorious for exaggerating network signal strength for example. (Easily googlable)
#XgFMost of the time I see that, its' related to crummy apps wasting battery/bloatware
#snarfedi can vouch for the N4 and most android phones lasting through the day with average usage
#snarfedalthough many of us are probably heavier than average users :P
#snarfedand for android having a power-aware sync platform for a long time now
squeakytoy joined the channel
#snarfedtantek, you're right about a lot of "feel good" displays, but the android power usage monitoring isn't that. it's the real deal. grab a friend's device and check out the screen XgF mentioned
#snarfed(google uses it internally for power profiling their own apps)
#XgFI call out Google Calendar specifically because it is *very* good also. As someone who's quite intimate with Android's SyncServices APIs, you can watch them making perfect use of them
#tanteksnarfed: Not only are the bigcos not doing it well (Apple is perhaps the most obv example) but even their crappy attempts are nearly complete silos (aside from some carddav/caldav support by iOS vs gmail services).
#tantekEveryone syncs in their vertical proprietary stack.
#tantekWith maybe a bit of bizdev across the bigcos (or a lot of custom coding to make specific implementations of *dav work, eg iOS has special code to make gmail work)
#tantekWhereas if we get federation right, it seems like we could implement syncing merely as a special case of federation across your own devices in particular.
#bearfederation allows for a lot of the simpler forms of OT to be used for sync and it's merge requirements
caseorganic and lukebrooker joined the channel
#tantekbear is there documentation anywhere of all the forms of OT and what use cases justify each one?
#bearI only know about the stuff google open-sourced with wave
#tantekYeah I've stayed away from wave which seemed totally overdesigned
#bearif the data being modelled is simple at it's core, then diff's should be able to handle most of it
#bearor deal with ACD actions at the smallest unit
#tantekAll the progress we've made with indieweb has been via simplifying to minimal use cases, implementing those, getting them workin cross-site, and then iterating.
#tantekAnd punting on all the a priori over thinking / over design that seems all too common in the federated / social web space.
tantek joined the channel
#tantek!tell benwerd ironic that that opendagenda article is on wired rather than on the author's own blog.