#snarfedaaronpk: do you use DataMapper.auto_migrate or auto_upgrade to create the wm.io schema? or did you originally do it by hand and it hasn't changed? or…?
#Pierre-OYes, typical spam, I think that the first spam on webmention will mean "it is popular"
brianloveswords joined the channel
#petermolnarPierre-O not webmention spam, local comment through webmention spam
#gRegor`Yeah, my site still accepts local comments - comment form at the bottom.
#petermolnarI've turned that off a while ago; yesterday the whole xmlrpc as well after a friends site I'm hosting collected 2k pingback spam within a week...
#gRegor`Since I fixed the Akismet plugin a couple weeks ago, this is the only one that's gotten through
#petermolnarmy personal experience: with the current releases of the webmention and the semantic linkbacks plugin I have not seen any issues in the past 3 months
barnabywalters joined the channel
#petermolnarGWG how does something get in the core btw?
#GWGpetermolnar, I would not mind getting some people who are involved heavily in the Wordpress community to give webmentions a try.
wolftune joined the channel
#barnabywalterstantek: actually I never fixed that problem on Taproot, only Shrewdness, by stripping the protocol off the URLs when using URLs as unique identifiers for things
#barnabywaltersneed to do the same in Taproot’s webmention handling code
brianloveswords joined the channel
#GWGpetermolnar, you recommended the apcu object cache plugin?
npdoty joined the channel
#petermolnarGWG maybe not here, but I do recommend it, yes
#aaronpkyeah, lots of other details mentioned there
#tantek.comedited /FreeMyOAuth (+19) "move common top 3 delegated OAuth providers to the top for easier findability" (view diff)
#tantekkeeps forgetting that as one of the first silo and POSSE documentation pages, the /Twitter page doesn't have the same level of regular structure as other silos
#LoqiOAuth is an authorization protocol created to replace the need for client applications of a service (e.g. a silo) to ask for your username and password to the service in order to access the service on your behalf http://indiewebcamp.com/oauth
#aaronpkso it fetches the bridgy source URL and bridgy makes the facebook API call
#snarfedaaronpk: i don't know much about sinatra, ruby, etc, but if you parallelize incoming requests decently, and don't block anything important on i/o, it may not be a problem
#aaronpkthat's not the problem. it's just synchronously processing each webmention including verifying the source URL links to target. so it has to make an outgoing HTTP request before it can respond
#aaronpkso while bridgy is making a request to wmio, wmio is then also making a request to bridgy
#snarfedyup. it's the common pattern, since most wm handlers process synchronously
#snarfedfwiw, one nice part about handling synchronously and parallelizing effectively (i/o etc) is that you can return the status in the response and still scale
#kylewmsnarfed: does GAE allow running multiple workers to process the task queue?
#kylewmso that you could be sending multiple mentions simultaneously, even if the receiver takes a while to respond
#aaronpkoh also... wmio doesn't have its own microformats parser right now so it's making *another* HTTP request to parse the microformats
#snarfedkylewm: yup. it does by default; i explicitly made the queue serial
#snarfedbridgy's outgoing queue rarely gets big, and wms don't really need to be too timely, so i optimized for being able to scale arbitrarily while keeping resource usage constant-ish
#kylewma heads up for anyone wanting to jump on #26 -- the parser service that webmention.io uses is doing a bit of interpretation too (authorship and whatnot), replacing it won't be as simple as just swapping out the parser
#kylewm(I stumbled on that trying to figure out why it didn't resolve relative paths in e.g., my author image)
#julian``anyone else ever thought about this feature in chrome? when typing an address youve been to it should give a visual indicator that youve visited this page before somewhere
#aaronpknot sure what you mean... I use chrome's auto-complete all the time to go back tgo pages i've visited
#julian``if you've visited multiple pages with the same starting characters eg a.com aa.com aaa.com and aab.com it only shows a certain amount so maybe itd show all except aab.com and then when you type in aab.com you have no visual indicator if youve been on here before just a small ux feature i think would be useful
#julian``actually in this example you would get a visual indicator, but if you're typing the end of a url eg url.com/blogpost1 11 111 etc you wouldnt see 112
caseorga_ and friedcell joined the channel
#gRegor`Chrome's auto complete will also match later parts in the URL. I've used it a handful of times when I remembered part of a URL but not the domain