The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2016-03-02

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 Grinnz_ jabberwok++
00:11 jabberwok cheers, all.  here's a virtual toast to documentation
00:13 taichi joined #mojo
00:15 jberger 🍻
00:15 taichi joined #mojo
00:31 jabberwok unicode fails me again  =(
00:32 jberger "\N{CLINKING BEER MUGS}"
00:32 jberger I just had this conversation with someone else
00:32 jberger install Symbola from here: http://users.teilar.gr/~g1951d/
00:33 jberger then restart your browser
00:33 jberger its a unicode-spanning font
00:33 jabberwok ah. i wonder whether it works with pidgin
00:33 Grinnz_ hm, is it monospace?
00:34 jabberwok gosh... it works
00:34 jberger not likely, but in my experience the system Does The Right Thing (TM) and uses it when no other font suffices
00:34 pink_mist I like Courier New for monospace unicode
00:35 pink_mist it's not all-encompassing, but it's got a lot
00:50 disputin joined #mojo
00:59 acamari joined #mojo
01:06 disputin joined #mojo
01:09 lluad joined #mojo
01:37 cpan_mojo Mojo-Weixin-1.0.5 by SJDY https://metacpan.org/release/SJDY/Mojo-Weixin-1.0.5
01:39 taichi joined #mojo
01:41 cpan_mojo Mojo-Webqq-1.7.3 by SJDY https://metacpan.org/release/SJDY/Mojo-Webqq-1.7.3
02:06 sri jberger: umm, did we really miss the latest web framework benchmark by one day?
02:08 sri we talked about it feb 24, and the latest results are dated feb 25 -.-
02:11 sri omg, some of the python and ruby apps are killing it this time
02:11 sri those results are unreal
02:20 sri sinatra with puma gets 80k rps where we get 26k rps
02:21 sri https://github.com/TechEmpower/FrameworkBenchmarks/tree/master/frameworks/Ruby/sinatra-sequel
02:22 sri i've never seen those kinds of results with mrb
02:22 sri umm, mri
02:23 sri hmm, 256 threads
02:23 sri jberger: maybe give it some more worker processes? :o
02:25 sri oh yea
02:25 sri they use 40 cores!!!
02:26 sri roflcopter
02:26 sri jberger set the mojolicious app to 8 workers
02:26 sri yea, that's a little underpowered
02:27 sri wish perl had a simple way to get the number of cores
02:29 sri jberger: i'd just use a cpan module to determine the number of cores, and set it dynamically
02:33 jberger We talked about it then, it still isn't merged
02:35 sri jberger: why do you set workers => 8 in app.pl, but nproc*2 in app.conf?
02:35 sri that seems rather counterproductive
02:36 jberger Is the in-app version a default?
02:36 sri it doesn't seem to do anything but get in the way
02:36 jberger I think it was a fallback when I was less sure about their loader framework
02:36 jberger Their tooling used to be really strange
02:36 sri the whole merge thing seems sketchy
02:37 sri wish i had access to a 40 core box
02:37 jberger It could almost certainly be simplified now
02:37 sri would be nice to optimize hypnotoad a little
02:38 jberger sri: see query
02:39 sri i just can't imagine that puma would be faster than a properly optimized hypnotoad
02:46 sri how do you even check cpu saturation on a box with 40 cores?
03:08 asarch joined #mojo
03:20 espen1 joined #mojo
03:31 sri bet there's a lot of stuff that can be optimized, like high/low watermarks for accepting connections in each worker
03:45 bpmedley sri , jberger : Can the number of cpus in /proc/cpuinfo be counted for the number of cores?
03:46 sri he has nproc
03:52 noganex_ joined #mojo
03:59 * sri puts it on the todo list https://github.com/kraih/mojo/issues/925
04:09 * sri added some load testing hints
04:15 espen1 joined #mojo
04:17 sri and we need a baseline
05:06 zivester joined #mojo
05:11 espen_ joined #mojo
05:13 inokenty-w joined #mojo
05:21 espen_ joined #mojo
05:26 sri hmm, perhaps a lower multi_accept default would result in more evenly distributed connections in a big worker pool
06:22 sue joined #mojo
06:32 good_news_everyon joined #mojo
06:32 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v218P
06:32 good_news_everyon mojo/master 912bfac Sebastian Riedel: accepted connections need to be handled separately
06:32 good_news_everyon left #mojo
06:35 sri when you set max_connections to 1, any backend connection (like a Mojo::UserAgent keep-alive connection) can prevent new conenctions from getting accepted
06:36 sri essentially locking up the worker
06:44 stryx` joined #mojo
06:50 dod joined #mojo
06:54 dod joined #mojo
07:08 good_news_everyon joined #mojo
07:08 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v210r
07:08 good_news_everyon mojo/master 7eb7346 Sebastian Riedel: there is no need for stop_gracefully to be slow
07:08 good_news_everyon left #mojo
07:14 sue joined #mojo
07:24 ashimema joined #mojo
07:27 espen_ joined #mojo
07:35 good_news_everyon joined #mojo
07:35 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v21zL
07:35 good_news_everyon mojo/master 34665f4 Sebastian Riedel: fix race condition in prefork test
07:35 good_news_everyon left #mojo
07:49 Vandal joined #mojo
08:24 Jonis I can never figure out where sri is at any given point
08:24 Jonis seems to have some sort of pan-global sleep schedule
08:27 AndrewIsh joined #mojo
08:41 trone joined #mojo
08:53 mdom Mhh, maybe it's nitpicking but setting $ua->ca to 0 disables hostname verification even if 0 is an existing file with certificates
08:54 mdom Is this intended or should Mojo::IOLoop::Client check for definedness?
09:00 sue joined #mojo
09:27 osfabibisi joined #mojo
10:03 Quai joined #mojo
10:14 Quai sri: ref. https://github.com/kraih/mojo/issues/925, hopes this helps; http://www.quai.io/mojo-bm.txt
10:14 Quai but now, lunch
10:19 ashimema totally off topic.. how do you guys go about finding and hiring good perl talent? We're about to start looking for another developer (one non-mojo web app, and one legacy web app to work on).. all work from home stuff. Any hints on where to start looking?
10:24 bpmedley ashimema : Did you ask your people if they know anyone?
10:24 ashimema not yet.. wasn't entirely sure who to ask yet ;)
10:25 bpmedley How many employees do you have?  I'd start there.. :)
10:26 bpmedley You're much more likely to get a cultural and team fit that way.
10:28 ashimema small company.. 4 devs, 16 employees total (with around a 1/3rd of those begin part time)
10:28 bpmedley Cool.  What do y'all do?
10:29 ashimema We're in the Library sector.. mostly work on Koha library system... open source ILS
10:29 ashimema https://www.ptfs-europe.com/ is us
10:31 ashimema Our biggest problem is we don't have huge amounts of money to throw around :(.  But it's a lovely group of people.. I love working here and have stuck with it long beyond I thought I would (and have been offered allot more money elsewhere) because they pay 'enough' and are such a nice team to work with.
10:31 ashimema but that's not somthing that's easy to sell to prospective employees
10:33 ashimema (although UK based mostly, we're open to any english speaking europeans too) - though I wonder hwo that will go with the bloomin UK leave europe vote coming up
10:34 bpmedley Oh wow.. I didn't realize a vote like that was taking place.
10:35 ashimema yeah..
10:35 ashimema Some time in June..
10:37 ashimema scary times for the UK.. I can see us getting royally stuffed if the vote goes the 'wrong' way (I'm a stay in myself)..
10:37 ashimema if the USA ends up with Trump and we end up out of the EU we're pretty stuffed me thinks
10:38 bpmedley I'm relocating to the ocean floor
10:41 ashimema you US?
10:41 bpmedley Yip
10:42 ashimema I take it from that comment you not a pro Trump man then.. I'm yet to meet one
10:45 bpmedley I do not support Trump.  Quite an odd situation here, very frustrating.
10:47 ashimema that's pretty much how every american I’ve spoken to says they feel.. a bit powerless and being taken along in the tide.
10:48 ashimema I'm obviosly talking to all the wrong americans ( or maybe the right one's ) to get any insight as to why one would want him in.
10:49 HtbaaPi is it possible to run morbo in perfork mode? Need to be able to execute multiple requests at the same time during dev
10:52 Shaeto joined #mojo
10:55 bpmedley HtbaaPi : https://github.com/kraih/mojo/blob/master/lib/Mojo/Server/Morbo.pm#L64  <-- I see fork management for one process
10:57 HtbaaPi bpmedley: so no go then. Alright, no biggy, I can work around it. It's only a small part of the app anyway
11:05 mishanti1 ashimema: I'm sort of "pro-Trump", but only because it would be funny to see what happens in the next episodes of "USA - The reality show". The previous episodes have been pretty entertaining. :)
11:05 ashimema haha
11:15 marcus srsly? You want that man with his finger on the button?
11:18 marcus I felt like John Oliver covered quite well why Trump is a scary scary idea.
11:30 kaare joined #mojo
11:35 noize joined #mojo
11:44 mishanti1 Omnipotence does not come with precidency, thankfully.
11:54 marcus mishanti1: the us president does control nuclear weapons tho.
12:09 asarch joined #mojo
12:11 sue joined #mojo
12:12 cosimo sri: I'm running some benchmarks with fast.pl and wrk. should I just copy/paste the results from wrk? or just req/s?
12:21 jberger I wish Jon Oliver had gone into Trump University more
12:24 cosimo highest results are ~48k req/s with -w 48 (this is a 48 cores system)
12:25 cosimo every 2 or 3 wrk runs I get connect and read socket errors
12:29 jberger Anyway this is the good news with regard to trump
12:29 jberger http://www.gallup.com/opinion/polling-matters/188936/trump-negative-image.aspx
13:00 marcus jberger: Yeah, I still feel like he would be no threat in the primaries.
13:00 marcus even if the republicans would be crazy enough to go with him
13:02 nic The Republicans absolutely must select Trump -- the Democrats need all the help they can get
13:02 Quai cosimo: I did it too, see link from around 11:14 :P
13:03 nic Let me know if it's helpful to run tests with more cores.  I _think_ I can go up to ~100 cores
13:04 cosimo Quai: good boy
13:05 nic cosimo did you tweak your kernel params to max your backlog queue?
13:05 cosimo Quai: did you up file descriptors limits before running the tests? that might explain the abundant socket errors
13:05 cosimo nic: yes, we have a custom config
13:07 mcsnolte joined #mojo
13:15 Quai cosimo: no, I did not. but it was limited to 100 connections, so it shouldnt be a problem. Or?
13:18 cosimo maybe
13:21 Quai I get the same socket errors message after upping the limit, but I do see a performance increase too
13:48 crux joined #mojo
13:55 neilhwatson joined #mojo
14:03 ramortegui joined #mojo
14:15 acamari joined #mojo
14:28 kes joined #mojo
14:43 Shaeto left #mojo
14:51 sue joined #mojo
14:59 fontardion joined #mojo
15:01 mdom I pushes a preliminary commit to https://github.com/mdom/mojo/commit/5675a. But one probably would check if tls_ca is defined and not the empty string... but maybe the real problem here is the overloading of the ca attribute to set the ca file and change hostname verification?
15:02 mdom s/pushes/pushed/
15:04 sri are we in trouble? https://github.com/kraih/mojo/issues/925
15:11 sri Quai/cosimo: thanks, that seems to confirm that we have scalability issues
15:12 sri my first guess would be uneven distribution of connections
15:13 sri -M 5 might fix that
15:13 sri maybe even lower... -M 2
15:21 sri it's interesting that results already get inconsistent at 8 workers
15:23 sri sadly there is not much more i can do for now
15:24 sri would be so much easier if i had a big box with lots of cores
15:33 jberger I have an 8 core you could use right now
15:33 jberger I can probably dredge up something more powerful relatively easily
15:41 sue joined #mojo
15:54 bd sri: i have two 2 cpu (xeon e5620 @2.4) with 8 cores each machines i could use
15:55 sri 8 might be too few
15:55 bd rest is in prod so can't use those
15:58 disputin joined #mojo
16:09 good_news_everyon joined #mojo
16:09 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2Dvo
16:09 good_news_everyon mojo/master 2a04ea8 Sebastian Riedel: stop_gracefully only has to wait for accepted connections
16:09 good_news_everyon left #mojo
16:12 sri ok, looks like it's not as simple as i thought https://groups.google.com/d/msg/mojolicious/DwtUU86vRAI/T_LnyVupDwAJ
16:12 sri someone will have to put serious work into that
16:12 sri not necessarily me
16:14 sri i think jberger might actually be in a much better position to fix it
16:16 jberger as I said I can probably find the hardware, but I don't consider myself an expert in this area
16:17 sri the key is to find out why load is so unevenly distributed
16:32 jnbek joined #mojo
16:38 disputin joined #mojo
16:42 lluad joined #mojo
16:45 good_news_everyon joined #mojo
16:45 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2Dm0
16:45 good_news_everyon mojo/master f2d44b1 Sebastian Riedel: test the connection limit
16:45 good_news_everyon left #mojo
16:45 disputin joined #mojo
16:58 bradjm joined #mojo
17:13 dod joined #mojo
17:23 jnbek joined #mojo
17:28 disputin joined #mojo
17:28 disputin joined #mojo
17:29 disputin joined #mojo
17:30 disputin joined #mojo
17:30 disputin joined #mojo
17:30 disputin joined #mojo
17:39 sue joined #mojo
17:42 good_news_everyon joined #mojo
17:42 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2DRm
17:42 good_news_everyon mojo/master e32c80b Sebastian Riedel: rename is_listening to is_accepting
17:42 good_news_everyon left #mojo
17:46 sri it's kind of fascinating how bad the prefork daemon scales with more than 8 workers in the default config
17:47 sri makes you wonder what kind of rps might be possible with optimizations
17:48 sri suddenly a ruby server at 80k rps while we were at 16k makes sense
18:00 disputin joined #mojo
18:00 disputin joined #mojo
18:10 disputin joined #mojo
18:15 sri aha, i think i see how puma was optimized there
18:16 sri they hardcoded the number of threads to 256, because that's the only concurrency value the framework benchmarks use
18:17 sri and then they started one worker per core, splittig up the 256 threads on those workers
18:18 disputin joined #mojo
18:18 sri seems reasonable, we could replicate that
18:20 sri yea, pretty sure the key here is to set -c just low enough to handle all connections concurrently
18:22 sri it's the same result for puma, if you incrase the threadpool size it decreases performance
18:22 sri yay, got it \o/
18:22 pink_mist \o/
18:23 sue joined #mojo
18:25 sri the lesson here is that you have to optimize specifically for the workload you're facing
18:26 sri the Mojo::IOLoop changes i made earlier should actually help a lot with this, since it allows for -c to be set more precisely
18:31 sri jberger: you should get the best performance by setting workers to nproc and clients to 256/nproc+1
18:32 sri and accepts to 0
18:34 * batman hate that he doesn't have time to run tests :(
18:34 batman I got 24 cores, but no cpu time :/
18:35 sri i wonder if the linux kernel version also matters
18:36 Adura Realtime kernel for the realtime web...!
18:36 sri since i heard that newer versions got better at load balancing when multiple processes are accepting from the same listen socket
18:36 batman sri: I can try to run some tests during daytime tomorrow, if that's not too late.
18:36 jnbek joined #mojo
18:36 sri batman: tests are always good
18:37 sri even if they just confirm what we already know
18:37 batman Got Centos 6.5 and 7. Not sure which version of kernel is currently running.
18:37 batman I think both servers has 24 cores - at least the 6.5 version has.
18:38 batman But does the load have to be 0 before starting the tests or can I get valuable data even if the system has *some* load?
18:38 jberger sri: I think the concurrency settings are available via the evironement in the test
18:38 jberger gotta look it up again
18:39 sri batman: 0 would be perfect, but a little load shouldn't matter too much (just make sure to mention it)
18:39 jnbek joined #mojo
18:39 sri jberger: the description says they only test with 256 now
18:39 batman Ok. Will try to minimize the load...
18:40 jberger well a lot of the tooling has changed since when I started working on these
18:40 batman I hope I don't get screwed over by a faulty production environment tomorrow :(
18:40 sri but "clients => ($CONCURRENCY / `nproc`) + 1" should be perfect
18:41 batman I'm very interested in this, since I was running benchmarks this Tuesday in production, and it suddenly went sideways.
18:41 pink_mist should get Mojolicious running on CUDA (or whatever is the popular way of using GPUs for computation these days) and try with thousands of cores :P
18:41 batman It wasn't mojo' fault though, but I would like to be more in control of the settings I tweak.
18:42 sri i was kinda hoping that there would be some magical pattern Mojo::IOLoop could use to scale automatically for every workload, but that just doesn't exist ;p
18:43 bradjm joined #mojo
18:43 sri you just have to optimize for every benchmark individually
18:43 batman sri: maybe not, but it would be awesome if the findings from this ended in a blog post/guide
18:44 batman ...to make it easier for others to tweak
18:44 batman I'm mostly just guessing :P
18:45 batman Quite silly. But it's hard when I try to make nginx, mojo, MySQL and redid to play along... Too many parameters to tweak :(
18:45 batman I think nginx is the only component I trust to "just work"
18:45 sri you should hire a consultant ;p
18:46 batman Indeed.
18:46 jnbek joined #mojo
18:46 batman Are you good at tweaking Linux, MySQL and nginx?
18:46 batman :)
18:47 renormalist joined #mojo
18:50 jnbek joined #mojo
18:51 jberger sri: http://frameworkbenchmarks.readthedocs.org/en/latest/Project-Information/Expected-Questions/#the-tests
18:52 jberger point #6
18:52 jnbek joined #mojo
19:03 mspo ou could detect how many cores are on the box
19:03 pink_mist that's what the `nproc` is for isn't it?
19:05 mspo you might also want to (on linux) explicitly tie a fork to its cpu or a smaller group of cpus
19:05 mspo like at most two
19:05 jberger sri: accepts should be 0? what about multi_accept?
19:06 sri making some changes to that in master right now ;p
19:07 disputin joined #mojo
19:07 rpaz joined #mojo
19:14 jberger sri: https://github.com/TechEmpower/FrameworkBenchmarks/blob/master/toolset/benchmark/framework_test.py#L176-L181
19:19 disputin joined #mojo
19:22 disputin joined #mojo
19:24 sri multi_accept is actually really problematic
19:25 sri because it doesn't take into account how many active connections we already have
19:28 disputin joined #mojo
19:38 renormalist joined #mojo
19:38 lluad joined #mojo
19:38 ramortegui joined #mojo
19:38 mcsnolte joined #mojo
19:38 asarch joined #mojo
19:38 Vandal joined #mojo
19:38 espen_ joined #mojo
19:38 ashimema joined #mojo
19:38 stryx` joined #mojo
19:38 inokenty joined #mojo
19:38 bpmedley joined #mojo
19:38 crab joined #mojo
19:38 jzawodn joined #mojo
19:38 twinhooker joined #mojo
19:38 romel joined #mojo
19:38 mdom joined #mojo
19:38 Mikey joined #mojo
19:38 batman joined #mojo
19:38 CHYC joined #mojo
19:38 bc547 joined #mojo
19:38 mgrimes joined #mojo
19:38 marcusr joined #mojo
19:38 vicash joined #mojo
19:38 Phil21 joined #mojo
19:38 dustinm joined #mojo
19:38 macaque joined #mojo
19:38 nicomen joined #mojo
19:38 isomorphismes joined #mojo
19:40 jberger NETSPLIT ALL THE THINGS!!!!
19:57 sri so, i want to turn multi_accept into a boolean
19:57 sri because i can easily check if the ioloop wants more connections
19:58 sri only problem is i have no idea how to name the new command line option
19:58 sri right now we have --multi-accept 10
19:59 sri and i want it activated by default
20:00 PryMar56 joined #mojo
20:27 sue joined #mojo
20:34 good_news_everyon joined #mojo
20:34 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2ytg
20:34 good_news_everyon mojo/master 8c39355 Sebastian Riedel: fix bug in Mojo::IOLoop where too many connections could be accepted
20:34 good_news_everyon left #mojo
20:34 sri this is the magic line https://github.com/kraih/mojo/commit/8c39355c597082d0b84d8ded19f477902b3e7ec7#diff-a9009b150b211dbbfc0a7453ca27e660R121
20:35 sri since we now have $self->{active} we know exactly when to stop
20:35 sri but configuring a number for --multi-accept doesn't make much sense anymore
20:36 sri (i think)
20:40 good_news_everyon joined #mojo
20:40 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2ymv
20:40 good_news_everyon mojo/master 4e71ed4 Sebastian Riedel: no need to link to Mojo::IOLoop
20:40 good_news_everyon left #mojo
20:42 sri hehe, only thing keeping me from making multi_accept a boolean is that i can't find a good name for a boolean command line option
20:42 sri --single-accept doesn't sound right
20:50 vicash --multi-accept 1 ?
20:51 sri that would be inconsistent with all other boolean flags
20:51 sri and i think you mean --multi-accept 0
20:51 sri since 1 is the default
20:51 vicash right. that would automatically mean single-accept or --no-multi-accept
20:53 sri --no-multi-accept is the best so far
20:59 mspo having a --no would mean you have a --multi-accept (like ./configure)
21:10 sri haha, maybe we just keep it the way it is ;p
21:11 sri maybe there's a little value in arbitrary numbers for testing
21:11 jberger I guess Getopt::Long doesn't make it easy for an option to have an optional value
21:12 jberger where passing a value would just get you a deprecation notice probably
21:12 ribasushi jberger: it... does (optional values)
21:12 jberger I was just looking for that
21:14 jberger sri: in the interim perhaps use 'multi-accept:i' and warn/ignore the value and add a 'no-multi-accept'
21:15 sri umm
21:15 jberger then after a deprecation period just make multi-accept a negatable boolean
21:15 sri how to deprecate it is not the problem
21:15 jberger oh
21:15 sri the problem is the name of the new option
21:15 jberger well with proper deprecation then isn't --(no-)multi-accept ok?
21:16 jberger 'multi-accept!'
21:17 cpan_mojo App-txtnix-0.04 by MDOM https://metacpan.org/release/MDOM/App-txtnix-0.04
21:17 good_news_everyon joined #mojo
21:17 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2yl9
21:17 good_news_everyon mojo/master 23a7393 Sebastian Riedel: mention how to disable multi_accept
21:17 good_news_everyon left #mojo
21:17 jberger wow this #id is not what I expected: https://metacpan.org/pod/Getopt::Long#pod
21:17 jberger but it works
21:17 sri that's what i want to know!
21:20 jberger what you really want is --no-multi-accept right?
21:20 jberger well, I guess you only said "the best so far"
21:20 jberger anyway, that would be 'multi-accept!'
21:20 jberger with a default of true
21:21 sri don't know what i want
21:21 sri so far keeping the old --multi-accept seems like the best option
21:21 sri a boolean option should be easier, not more complicated
21:21 sri that's the whole point ;p
21:23 sri too bad they don't make laptops with more than 4 cores yet
21:29 mspo make a vm with more?
21:33 jberger sri: ping to #mojo-core (if you didn't see it)
21:39 PryMar56 joined #mojo
22:00 good_news_everyon joined #mojo
22:00 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2yaH
22:00 good_news_everyon mojo/master ac64ed3 Sebastian Riedel: use the same example everywhere
22:00 good_news_everyon left #mojo
22:05 good_news_everyon joined #mojo
22:05 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2yV5
22:05 good_news_everyon mojo/master b05ba01 Sebastian Riedel: a few more commas
22:05 good_news_everyon left #mojo
22:23 good_news_everyon joined #mojo
22:23 good_news_everyon [mojo] kraih tagged v6.52 at 710d6b7: https://git.io/v2yK5
22:23 good_news_everyon left #mojo
22:25 good_news_everyon joined #mojo
22:25 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/v2y6G
22:25 good_news_everyon mojo/master 5097e10 Sebastian Riedel: bump version
22:25 good_news_everyon left #mojo
22:30 sri looks like someone is sabotaging cpanratings again
22:30 sri hidden malicious ratings are popping up again :( http://cpanratings.perl.org/dist/Mojolicious
22:31 sri my all time favorite too... http://cpanratings.perl.org/dist/Mojolicious#8130
22:32 sri cpanratings needs to die
22:35 Adura Not everyone is going to like you.
22:36 sri exactly, and those people can very easily manipulate cpanratings
22:37 sri it's terrible
22:38 Adura Wonder how much that guy that email spams rt.cpan.org users about his module issues gets shit on
22:43 Grinnz_ http://cpanratings.perl.org/dist/Coro
22:43 Grinnz_ that guy? :)
22:43 Adura That guy, yes.
22:44 Grinnz_ i see 2 positive reviews, and a negative review that gave a 4 star rating for some reason
22:44 Adura Maybe they thought the stars were bad.
22:44 pink_mist maybe they thought the stars implied how many throwing stars they'd like to throw at the author?
22:46 Adura http://cpanratings.perl.org/user/mlehmann Doesn't work, unfortunately. So you can't separate modules with single authors from ones with multiple.
22:46 sri well, so i did this https://github.com/CPAN-API/metacpan-web/issues/1653
22:46 sri cpanratings must die
22:47 punter joined #mojo
22:48 jberger Commented
22:48 sri jberger++
22:57 sue joined #mojo
23:00 sri fighting with memes! https://twitter.com/kraih/status/705165560770859008
23:13 sri hahaha... omg i'm so tempted to answer with this http://i2.kym-cdn.com/photos/images/list/000/715/142/f6c.jpg
23:14 sri must... resist...
23:15 sri oh, i guess oalders already saw -.-
23:17 sri but yea, it's dead https://github.com/perlorg/perlweb/tree/master/lib
23:25 oalders yeah, CPAN Ratings does have its problems, but there are still helpful reviews on there
23:25 oalders i have no idea what would be involved in updating it, though
23:26 sri it would be less of a problem if you don't show the stars so prominently i suppose
23:27 oalders if there's a consensus that it should be removed, i'll go along with it.
23:28 oalders looks like it's on the Combust framework, which i've never worked with
23:32 jberger I see traces of it on cpan but I don't see the framework itself
23:32 jberger Bundle::Combust exists
23:33 meredith i think there are def some babies out there in the perl world but i also don't mind the idea of having a place where folks can get guidance out of the ratings we collectively post :\
23:34 meredith also mojolicious in particular seems to have some folks who are still inexplicably salty about it existing
23:35 jberger unicode fail: http://combust.develooper.com/install.html#Copyright
23:35 Grinnz_ heh
23:36 Grinnz_ "Ask Bj&#195;&#184;rn"
23:36 Grinnz_ it's not even a charset issue lol
23:36 Grinnz_ They somehow html-encoded the mojibake
23:39 sri quick summary of the 20+ core issue https://github.com/kraih/mojo/issues/925#issuecomment-191494560
23:41 sri if there's really doubt about the value of cpan ratings, i'll see if i can make a script for manipulating it next weekend
23:54 sue joined #mojo

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary