sbeam_ has quit [Remote host closed the connection]
havenwood has quit [Remote host closed the connection]
<indirect>
rubygems.org :(
mp___ has joined #rubygems
<mp___>
I'm finding Rubygems 2.2 dramatically slower for installing gems than 2.0 was. Filing an issue with traces now, but wondering if anybody else has had this problem
<Jamo>
mp___: have you tried installing ruby via rbenv or rvm?
<Jamo>
dunno, but it might have a difference...
<mp___>
Jamo: I found similar speed problems when installing 2.1 on my mac, which is via rbenv
<Jamo>
ok. (i havent yet tried 2.1.0 on my mac)
<mp___>
wish there was a way to go straight to the au-m mirror and avoid the api.rubygems.org bit
josh-k has joined #rubygems
<vertis>
indirect: is there a problem?
lsegal has quit [Read error: Connection reset by peer]
lsegal has joined #rubygems
<Jamo>
try /usr/bin/time -v or /usr/bin/time -l instead of shells own time command
<Jamo>
you should get more input
<Jamo>
* details
<Jamo>
mp___: ^
<mp___>
ta, will do
redmenace has joined #rubygems
diegoviola has joined #rubygems
<diegoviola>
hi
<diegoviola>
i read a comment on twitter saying "if you have rubygems 2.2.0 you don't need bundler", i'm not sure if i understood well, but rubygems 2.2 makes bundler obsolete?
<Jamo>
and its getting late, so ill probably fall asleep soon... :)
<mp___>
Jamo: ah, but did you have a clean gem dir?
<Jamo>
good point, ill do it again
<mp___>
Jamo: make sure there's no cached specs at all
<mp___>
Jamo: I was doing this on a scratch box
<mp___>
eg, still takes 3 minutes on a fresh install from us-east-1. two minutes form us-west-2 (same location as api.rubygems)
<Jamo>
actually, ill try this tomorrow mornig in clean vagrant box
<Jamo>
(its 5am... :D )
<Jamo>
(ill update the same gist
<mp___>
ta :)
<mpapis>
hmm, let me try this too
<mp___>
my ruby was built via a standard 'make install' as part of fpm-recipes
<mp___>
so .. hopefully unlikely that it's something wrong with my ruby build
<mpapis>
mp___, RG-2.1.11 is the same as 2.2.0 and 2.0.14 is a lot faster (3s vs 16s on gem install rvm)
<mp___>
ty
<mp___>
there seems to be a large latency component, and I see about 50% more GET requests on 2.2 - looks like it's trying to pull down gemspecs for other architectures
<mp___>
but even removing the latency component it's still slower
<mp___>
see issue - I've done tests from us-west-2 & us-east-1
<mp___>
I was worried that this was another "everybody who hacks on rubygems/bundler lives in the valley" problem ;)
<mpapis>
mp___, I'm testing down 2.1.11 - 2.1.0 to find when it happened
<diegoviola>
i did "gem install -g" and that worked well, it installed all the gems from my Gemfile, but when i run the same command again, i get this: "ERROR: While executing gem ... (ArgumentError) Malformed version number string 3.16.14.3-x86_64-linux"
<mpapis>
diegoviola, open a bug report with your Gemfile + Gemile.lock and all the commands + outputs
<diegoviola>
ok
<pipecloud>
Here's something I mentioned early this morning, but didn't get any feedback on. I'm a bit stuck on trying to get rubygems.org to run on ruby 2.0 or 2.1. https://gist.github.com/Spaceghost/8214607
<mpapis>
mp___, I can not get consistent results I get someimes ~6s sometimes ~16s the same ruby and RG
<mp___>
mpapis: :(
<mp___>
mpapis: interesting. if I get time I might run some more tests
<mp___>
mpapis: avgtime might be useful here
<pipecloud>
I'm working on fixing it. I was at xavier shay's talk and we chatted shortly about this, so I was going to give it a go.
<mpapis>
mp___, the same ruby and RG just trying between different versions of RG and I get 10s difference in one RG version, I do not see anything common, make sure to pick light gem like "rvm" to limit network and cpu/size issues
<mpapis>
pipecloud, your message lacks context for me
<mp___>
mpapis: Thing is I think it'll be most marked with a gem with lots of deps, like chef
seanlinsley has quit [Quit: …]
<pipecloud>
mpapis: I'm using ruby 2.1, I've cloned rubygems.org, bundle'd with bundler from HEAD, I've got rubygems, and I've checked out both HEAD and https://github.com/square/rubygems/commit/f24c244308266970fdfdee85f5cba8532c936b79 because xshay said it was necessary. I'm running tests and test/unit/pusher_test.rb has three breakages.
<mpapis>
mp___, my point is if I can not get consistent results on small one how do you want to compare big ones ;)
<mp___>
mpapis: oh! yes.
<pipecloud>
mpapis: One of which is that the Pusher class is receiving #new with a StringIO file as the second argument.
<mpapis>
mp___, also 2.2 has reworked dependency resolver with support for Gemfile.lock - maybe that's your issue
<mp___>
mpapis: I suspect that hte problem is in the dependency resolution, so the problem will only reliably manifest with a gem with lots of deps
<mp___>
mpapis: also - if I run it ten times on RG 2.1.11 and get different results for each run, but run it ten times on RG 2.0.0 and get the same result each run ..
<pipecloud>
mpapis: That StringIO is from a gem archive that sj26 added a while ago. They're in test/gems/dos* They aren't parseable, and ruby shits itself saying that string contains null byte if I call #read to pass an actual string to Pusher.new instead of passing in the StringIO object.
<mp___>
I think those of us on high-latency connections would see a big speedup if rubygems didn't insist on redirecting to mirrors for tiny specfiles
<pipecloud>
mpapis: Would it help if I linked directly to the code I'm referencing?
<mpapis>
pipecloud, I could have a look, but I'm only helping here, not a real expert ;)
<pipecloud>
mpapis: Thanks. I'll give that a go. It explains why doing a similar change on the test_helper method yielded no good results.
<mpapis>
not sure why you would use StringIO.new(body.read) instead of just talking to the body directly
<pipecloud>
I don't know just yet, I'll look into it and see if I can figure it out.
<diegoviola>
got another issue with gem install -g, i have this line in my Gemfile: "gem 'pry-rails', group: :development", however, after installing gems with gem install -g, rails c won't even start, it will say "Could not find gem 'rr (>= 0) ruby', which is required by gem 'pry-rails (>= 0) ruby', in any of the sources."
<diegoviola>
some gems that are needed by pry-rails are not installed, rr and appraisal
<diegoviola>
if i git checkout Gemfile.lock; rm -rf ~/.gem and re-install things with bundler, it will install everything
<diegoviola>
and rails c will work
<Rennex>
diegoviola: does gem install pry-rails without the -g option install its dependencies?
<diegoviola>
Rennex: hrm, let me try that
<diegoviola>
Rennex: no, it won't install rr or appraisal
<diegoviola>
not sure what gem actually gets those
<Rennex>
diegoviola: those seem to be development dependencies
<Rennex>
but i have no clue how gem without bundler deals with development and production environments :)
<diegoviola>
does gem install -g deals with those diffrently than bundler?
<Rennex>
i think bundler defaults to development if it's not set
<diegoviola>
hrm
<diegoviola>
thanks
redmenace has joined #rubygems
<Rennex>
i bet someone here knows how to make gem behave...
<mpapis>
diegoviola, you still have to report it - in that case for bundler that it installs gem.add_development_dependency on bundle install of gems group: development
<mpapis>
I do not think it should happen
<diegoviola>
mpapis: so it could be a bug?
<mpapis>
or for RG to make sure it should not happen
<mpapis>
diegoviola, accidental problem with bundler and rubygems not doing that - because it is wrong ... at least it is my opinion
<mpapis>
maybe open it for RG and you will get confirmation what is the right thing to happen
<diegoviola>
RG?
<mpapis>
rubygems
<diegoviola>
ok
<mpapis>
to long to type for me --- sorry
<diegoviola>
i'll open this issue tomorrow first thing in the morning
<diegoviola>
np
redmenace has quit [Ping timeout: 252 seconds]
<diegoviola>
another thing that kind of bothered me about 'gem install -g' is that i started a new rails app, ran 'gem install -g', installed all gems ok, ran it again, and i think it tried to install everything again
<diegoviola>
i think it should check that i already installed things and not do anything
<diegoviola>
bundler at least behaves like that
<diegoviola>
i'm not sure if it was reinstalling everything though
<diegoviola>
but it was installing things the second time i ran it
<diegoviola>
anyway
<diegoviola>
i'm crashing to bed now, bbl :p
<diegoviola>
thanks for all your help, mpapis and Rennex
diegoviola has quit [Quit: WeeChat 0.4.2]
mp___ has quit [Remote host closed the connection]
josh-k has quit [Remote host closed the connection]
josh-k has joined #rubygems
vertis has quit [Quit: vertis]
josh-k has quit [Ping timeout: 272 seconds]
arohner has quit [Remote host closed the connection]
dogweather has joined #rubygems
redmenace has joined #rubygems
redmenace has quit [Ping timeout: 272 seconds]
arohner has joined #rubygems
tbuehlmann has joined #rubygems
arohner has quit [Ping timeout: 260 seconds]
arohner has joined #rubygems
arohner has quit [Ping timeout: 245 seconds]
lsegal has quit [Quit: Quit: Quit: Quit: Stack Overflow.]
redmenace has joined #rubygems
redmenace has quit [Ping timeout: 272 seconds]
arohner has joined #rubygems
arohner has quit [Ping timeout: 245 seconds]
josh-k has joined #rubygems
josh-k_ has joined #rubygems
josh-k has quit [Ping timeout: 272 seconds]
wizonesolutions has quit [Ping timeout: 260 seconds]
wizonesolutions has joined #rubygems
workmad3 has joined #rubygems
redmenace has joined #rubygems
redmenace has quit [Ping timeout: 252 seconds]
workmad3 has quit [Ping timeout: 272 seconds]
arohner has joined #rubygems
dbussink_ has quit [Remote host closed the connection]
dbussink has joined #rubygems
arohner has quit [Ping timeout: 245 seconds]
josh-k has joined #rubygems
josh-k_ has quit [Ping timeout: 252 seconds]
josh-k_ has joined #rubygems
josh-k has quit [Ping timeout: 260 seconds]
arohner has joined #rubygems
arohner has quit [Ping timeout: 252 seconds]
huoxito has joined #rubygems
redmenace has joined #rubygems
redmenace has quit [Ping timeout: 252 seconds]
josh-k has joined #rubygems
josh-k_ has quit [Ping timeout: 272 seconds]
diegoviola has joined #rubygems
josh-k has quit [Read error: Connection reset by peer]
josh-k has joined #rubygems
<diegoviola>
hi
<diegoviola>
so i'm trying to figure out why 'gem install -g' won't install things like rr and appraisal, but it does when i run bundle install
<diegoviola>
any ideas?
<diegoviola>
should i file a bug about this?
bbrowning has joined #rubygems
josh-k has quit [Remote host closed the connection]
seanlinsley has joined #rubygems
<mpapis>
diegoviola, yes please, always open a ticket when in doubt ;) ... at least for me
<diegoviola>
ok, but i'm not sure it's actually an issue or it's just pebkac here
<diegoviola>
i just created a new rails app, i called it 'foo', then added "gem 'pry-rails', group: :development'
<jeremyevans>
drbrain: Not sure if you are back from vacation yet, but if so I'd like to know if there is a way to configure rubygems 2.2.0 to build extensions in the source directory (instead of the new extensions/$arch/$ruby_version/$gem_name_and_version directory)
<jeremyevans>
drbrain: Basically, build extensions similar to previous versions of rubygems
<drbrain>
I am back!
<jeremyevans>
great!
<drbrain>
why do you want it to work the old way?
<drbrain>
just so I understand
<jeremyevans>
For OpenBSD packaging, if everything builds in a new directory, it's very difficult to get the same port to build packages for multiple ruby versions
<jeremyevans>
Well, not very difficult, but not simple and it requires a nontrivial amount of work
<drbrain>
how do you build the packages now?
<drbrain>
`gem install`?
<mpapis>
jeremyevans, now you can compile one gem for multiple ruby versions - simpler?
<jeremyevans>
Pretty much
<jeremyevans>
mpapis: No, because the packages are separate (ruby19-foo for ruby 1.9, ruby20-foo for ruby 2.0, etc.)
<mbklein>
drbrain: As long as you're around answering questions... :) ... any idea why this gem refused to push to rubygems ("There was a problem saving your gem: Home does not appear to be a valid URL") until I commented out the homepage attribute? https://github.com/mbklein/handle/blob/master/handle.gemspec
<mpapis>
jeremyevans, you could merge now and save amount of packages and space
redmenace has joined #rubygems
<jeremyevans>
mpapis: Well, it can't work that way for older versions of ruby, which still need to be supported
<drbrain>
jeremyevans: in that case, I could probably add an option to install to do it the old way, would that work?
<drbrain>
jeremyevans: for now, extensions get built to the new directory and to lib/ like rubygems 2.1.x and earlier
<jeremyevans>
drbrain: I think that would work. But is there any chance of it getting backported to 2.1?
<drbrain>
so you could just throw away the extensions/ directory
<drbrain>
jeremyevans: sure, I could backport it
<drbrain>
mbklein: there's only one slash following the scheme
<jeremyevans>
drbrain: That's basically what my workaround will do, though there are issues if you don't have the build_complete file
<mbklein>
lolwut
<mbklein>
drbrain: I swear I looked at that line SO MANY TIMES. How embarrassing. :p
<drbrain>
jeremyevans: ah, hrm
<drbrain>
mbklein: fresh eyes :D
alindeman has quit [Ping timeout: 240 seconds]
<jeremyevans>
drbrain: Currently, since the packages are installed by root, the user doesn't have the ability to build extensions, and you get an permissions error when trying to load the gem
alindeman has joined #rubygems
vertis has joined #rubygems
* drbrain
nods
<jeremyevans>
I'm guessing I can just touch that file to work around the issue, is that correct?
vertis has quit [Client Quit]
<drbrain>
yes
<jeremyevans>
That's basically what I planned to do if the option couldn't be added
<drbrain>
can you create an issue for me?
<jeremyevans>
OK. Thanks for the help!
vertis has joined #rubygems
redmenace has quit [Ping timeout: 264 seconds]
graphex has quit [Ping timeout: 248 seconds]
<drbrain>
np! all the ways rubygems is used is unknowable to me, so I'm happy to (if partially) restore old behavior I didn't know existed
havenwood has quit [Remote host closed the connection]
havenwood has joined #rubygems
<jeremyevans>
drbrain: Issue submitted. Thanks again for your help. :)
<drbrain>
:D
<jeremyevans>
drbrain: Another thing I noticed is that the extensions directory uses RUBY_VERSION instead of RbConfig::CONFIG['ruby_version'], which leads to a weird name if you configure ruby with --with-ruby-version=minor
<drbrain>
ok, can you create an issue for that too?
<drbrain>
I thought I had it using the "ABI version"
<jeremyevans>
Sure, I'll add an issue for that as well
dvu has joined #rubygems
almostwhitehat has quit [Remote host closed the connection]
<jeremyevans>
drbrain: Looks like the X.Y.Z version is hard coded in Gem.ruby_api_version. Should that just be changed to RbConfig::CONFIG['ruby_version'], or should it be changed at another level?
<drbrain>
ruby_api_version is "#{ConfigMap[:MAJOR]}.#{ConfigMap[:MINOR]}.#{ConfigMap[:TEENY]}"
<drbrain>
which is RbConfig::CONFIG['MAJOR'], etc.
<drbrain>
is it the same?
bbrowning is now known as bbrowning_away
<jeremyevans>
I'm sure it's the same information. It looks like Gem::ConfigMap[:ruby_version] will work, but again, I'm not sure if ruby_api_version is used in other places that could cause such a change to break things.
<drbrain>
looks like they're the same in rbconfig for 1.9.3 (ruby_version of 1.9.1)
diegoviola has quit [Quit: WeeChat 0.4.2]
<jeremyevans>
drbrain: I added an issue for changing Gem.ruby_api_version, since I think that's probably the correct place
<drbrain>
thanks!
seanlinsley has quit [Quit: …]
davidjrice has quit [Ping timeout: 272 seconds]
ckrailo has quit [Ping timeout: 246 seconds]
davidjrice has joined #rubygems
avdi has quit [Ping timeout: 246 seconds]
nz has quit [Ping timeout: 246 seconds]
JSharp has quit [Ping timeout: 272 seconds]
xerxas has quit [Ping timeout: 272 seconds]
ckrailo_ has joined #rubygems
nz_ has joined #rubygems
johnmwilliams___ has quit [Ping timeout: 272 seconds]
pipecloud has quit [Ping timeout: 246 seconds]
JSharp has joined #rubygems
seanlinsley has joined #rubygems
gazoombo has quit [Ping timeout: 240 seconds]
davidjrice has quit [Ping timeout: 246 seconds]
ckrailo_ has quit [Ping timeout: 246 seconds]
guilleiguaran has quit [Ping timeout: 246 seconds]
nz_ has quit [Ping timeout: 240 seconds]
ckrailo_ has joined #rubygems
nz_ has joined #rubygems
guilleiguaran has joined #rubygems
JSharp has quit [Ping timeout: 264 seconds]
guilleiguaran has quit [Ping timeout: 264 seconds]
guilleiguaran has joined #rubygems
tmoore has quit [Ping timeout: 246 seconds]
tmoore has joined #rubygems
guilleiguaran has quit [Ping timeout: 246 seconds]
guilleiguaran has joined #rubygems
workmad3 has joined #rubygems
ckrailo_ has quit [Ping timeout: 245 seconds]
guilleiguaran has quit [Ping timeout: 246 seconds]
mbklein has quit [Ping timeout: 245 seconds]
ckrailo_ has joined #rubygems
Guest85414______ has quit [Ping timeout: 246 seconds]