|
Jonny 290 posted:we were laughing about it in irc last night, basically the two walmarts in town are the most dispatched addresses by a factor of two. the third is the police station itself "hey guys, come back to the station; elmer's gotten into the whiskey again"
|
# ? Jun 20, 2013 19:04 |
|
|
# ? May 8, 2024 06:25 |
|
a lot of them are rape charges they don't dispatch those to the real address, they put out the call and then the cops dial in on their cell phone btw if yall didnt know cops do a hilarious amount of off the record communication with HQ on their cell phones, it's why they don't push for expensive encrypted poo poo more i have the duty cell number of almost every cop in town because they constantly have to remind the dispatcher
|
# ? Jun 20, 2013 19:08 |
|
Otto Skorzeny posted:Look through your last few months of logs, you will probably find that addresses follow an exponential distribution or something like it. Scrape google once for the top 100 or whatever and throw em in a lookup table. automating this sort of thing only gets you 90% there anyway. unless you do some curating the results will always be a little off
|
# ? Jun 20, 2013 19:09 |
|
Jonny 290 posted:a lot of them are rape charges they don't dispatch those to the real address, they put out the call and then the cops dial in on their cell phone well that took the starch out of my attempted joke people suck (not you, jonny, the people the cops are after)
|
# ? Jun 20, 2013 19:39 |
|
Nomnom Cookie posted:automating this sort of thing only gets you 90% there anyway. unless you do some curating the results will always be a little off If you actually use a geocoding API instead of scraping they usually specify a confidence value for the result. They curate stuff themselves so there's no reason for u to duplicate the work.
|
# ? Jun 20, 2013 20:06 |
|
i'll have to investigate right now i have ghetto lat/long decoding going on from geocoder.us or something like that, i'll have to dig deeper
|
# ? Jun 20, 2013 20:14 |
|
what is Geo::Coder::Googlev3 like
|
# ? Jun 20, 2013 20:15 |
|
Jonny 290 posted:yeah that will prob work tbqh matlab
|
# ? Jun 20, 2013 20:17 |
|
Bloody posted:matlab no don't
|
# ? Jun 20, 2013 21:33 |
|
Bloody posted:matlab pppppppppffffffffffbbbbbtttttttttttttttttttttttttttttttttttttttttttt use r
|
# ? Jun 20, 2013 21:42 |
|
don't ever follow my posting advice
|
# ? Jun 20, 2013 21:43 |
|
aaagh unicode http://labs.spotify.com/2013/06/18/creative-usernames/
|
# ? Jun 20, 2013 22:33 |
|
Solution: don't canonicalise, only refer to users by guid
|
# ? Jun 20, 2013 23:44 |
|
I am not a guid, I'm a free man.
|
# ? Jun 20, 2013 23:45 |
|
Zombywuf posted:Solution: don't canonicalise, only refer to users by guid i see gold fringe around this post
|
# ? Jun 20, 2013 23:46 |
|
Zombywuf posted:Solution: don't canonicalise, only refer to users by guid Agreed but don't you still want to canonicalise to make sure "duplicated" usernames aren't signing up? I imagine that would lead to a lot of confusion with like the Omega and Ohm symbols and passing yourself off as another user with a similar looking name.
|
# ? Jun 21, 2013 01:56 |
|
As mentioned in the article, the trick is to either restrict the input to valid unicode 3.2 or to find a fixed-point in your canonization (or reject entries that do not find one quick enough). It's a real nice lovely thing to realize and find about though.
|
# ? Jun 21, 2013 01:58 |
|
The best part of the article was that upgrading form python 2.4 to 2.5 broke it LOL
|
# ? Jun 21, 2013 03:45 |
|
prefect posted:i see gold fringe around this post new james bond movie should be Goldfringer, about a villain that aims to control the world via obscure admiralty law loopholes
|
# ? Jun 21, 2013 03:51 |
|
Jonny 290 posted:new james bond movie should be Goldfringer, about a villain that aims to control the world via obscure admiralty law loopholes this would work better if fringe wasn't pronounced with a soft g like gif
|
# ? Jun 21, 2013 03:56 |
|
MononcQc posted:As mentioned in the article, the trick is to restrict the input to ASCII
|
# ? Jun 21, 2013 04:36 |
|
Zombywuf posted:Solution: don't canonicalise, only refer to users by guid i am not a number i am a free man!
|
# ? Jun 21, 2013 09:12 |
|
use email / password for login dont let users pick names that look like other usernames (render them to images and do a edit distance comparison)
|
# ? Jun 21, 2013 10:12 |
|
design your system so that what passes for usernames can change heck, design so anything that can identify a user can change later Carthag posted:(render them to images and do a edit distance comparison) woah but this is likely not as easy as it sounds
|
# ? Jun 21, 2013 13:53 |
|
Using e-mails as log-in names may work, but you want to provide some other immutable underlying ID so that a user can change e-mails without a problem. You'd probably end up with a general canonical unique ID, a primary e-mail (that can be verified and used to log in), and a 'display name' that may or may not be unique following canonicalisation as requirements arise.
|
# ? Jun 21, 2013 14:04 |
|
Also, distributed computing link of the day, this time on CRDTs: http://hal.upmc.fr/docs/00/55/55/88/PDF/techreport.pdf More (shorter): http://pagesperso-systeme.lip6.fr/Marc.Shapiro/papers/RR-6956.pdf MononcQc fucked around with this message at 14:27 on Jun 21, 2013 |
# ? Jun 21, 2013 14:05 |
|
Email is a very good username cause its both unique and the domain part can be used to determine organizational specific parameters. Ex: using federated login for certain orgs. So really your authentication system should be wholly separate from authorization. Use claims/assertion based authorization.
|
# ? Jun 21, 2013 14:24 |
|
Hard NOP Life posted:Agreed but don't you still want to canonicalise to make sure "duplicated" usernames aren't signing up? I imagine that would lead to a lot of confusion with like the Omega and Ohm symbols and passing yourself off as another user with a similar looking name. When I said "refer to" I meant in the UI.
|
# ? Jun 21, 2013 14:50 |
|
qntm posted:i am not a number yessssssssssssssss https://www.youtube.com/watch?v=OuDVXh_a1-Y https://www.youtube.com/watch?v=a7LH__BPqSY
|
# ? Jun 21, 2013 15:10 |
|
Otto Skorzeny posted:yessssssssssssssss best part is the laugh number two thinks it's absolutely hysterical that a man could be free
|
# ? Jun 21, 2013 17:44 |
|
Max Facetime posted:woah one man's ﷽ is another's man's □
|
# ? Jun 21, 2013 19:11 |
|
|
# ? Jun 21, 2013 19:22 |
|
b0lt posted:one man's ﷽ is another's man's □ b0lt did 9/11
|
# ? Jun 21, 2013 19:30 |
|
signing up for spotify as '﷽ 卐卐卐卐卐 OSAMA BIN HITLER 卍卍卍卍卍 ﷽<right to left override mark>' as we speak
|
# ? Jun 21, 2013 20:48 |
|
b0lt posted:one man's ﷽ is another's man's
|
# ? Jun 21, 2013 21:29 |
|
not quite PL but im a rebel Are Subqueries RESTful? well tef. are they tbh i didnt read the article cause it looks boring and idgaf. i just hope its terrible and pisses someone off
|
# ? Jun 23, 2013 08:27 |
|
Urgh, subqueries are irrelevant to REST. They're just other resources identified by a URL which the resource representation's you've already received should link to (or otherwise specify how to construct).
|
# ? Jun 23, 2013 12:26 |
|
yeah but are they a monoid?
|
# ? Jun 23, 2013 12:42 |
|
AlsoD posted:yeah but are they a monoid? don't use ableist language for shame
|
# ? Jun 23, 2013 12:44 |
|
|
# ? May 8, 2024 06:25 |
|
AlsoD posted:yeah but are they a monoid? Nearly everything is a monoid. If not, just consider its endomorphism monoid.
|
# ? Jun 23, 2013 12:46 |