Ruby on Rails Love-In

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Ruby on Rails Love-In

«‹›178 »

Pardot: Jul 25, 2001

BonzoESC posted:

How did I make rake assets:precompile depend on git?

Something in your precompile is calling out to the git command. To simulate it locally, I'd copy your project to a new directory and rm the .git directory and try RAILS_GROUPS=assets RAILS_ENV=production bundle exec rake assets:precompile

# ? Oct 31, 2011 00:14

Adbot: ADBOT LOVES YOU

# ? May 14, 2024 06:37

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

Looks like I'm running in to https://github.com/rails/rails/issues/2765 (the image not being precompiled part, the git part was a red herring).

In config/environments/production.rb:

code:

53   # Don't fallback to assets pipeline
54		-  config.assets.compile = false
54		+  config.assets.compile = true

Also, did the message just change to include the rake command, or does it not show up on failure?

code:

-----> Preparing app for Rails asset pipeline
       Running: rake assets:precompile
       fatal: Not a git repository (or any of the parent directories): .git
       /usr/local/bin/ruby /tmp/build_1fcg6w8ore77/vendor/bundle/ruby/1.9.1/bin/rake assets:precompile:nondigest RAILS_ENV=production RAILS_GROUPS=assets
       fatal: Not a git repository (or any of the parent directories): .git

# ? Oct 31, 2011 00:38

Pardot: Jul 25, 2001

BonzoESC posted:

Also, did the message just change to include the rake command, or does it not show up on failure?

We do show you both stdout and stderr, but "nondigest" is nowhere in the code for what we do on slug compile. It must be from what your (or rails') rake task is doing.

If you're still having a lot of problems with the assets stuff, you should open a support ticket. The support guys have a lot more experience troubleshooting this than I do.

# ? Oct 31, 2011 01:41

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

Pardot posted:

We do show you both stdout and stderr, but "nondigest" is nowhere in the code for what we do on slug compile. It must be from what your (or rails') rake task is doing.

If you're still having a lot of problems with the assets stuff, you should open a support ticket. The support guys have a lot more experience troubleshooting this than I do.

Fun stuff; I wonder if it's compass, sass-rails, kumade, or any of the million other gems. It does work though.

# ? Oct 31, 2011 02:25

NotShadowStar: Sep 20, 2000

The asset pipeline is a horribly implemented idea and I hope it either goes away or gets fixed. I keep running into these horror stories repeatedly about it just not working right.

# ? Oct 31, 2011 04:26

Nybble: Jun 28, 2008; praise chuck, raise heck

Ruby 1.9.3 is out! http://news.ycombinator.com/item?id=3174477

Looks like it's a lot faster by fixing some thread issues. http://www.rubyinside.com/ruby-1-9-3-faster-loading-times-require-4927.html

For those using RVM, it's easy to update:

code:

rvm get head
rvm reload
rvm install 1.9.3
rvm use 1.9.3 --default

# ? Oct 31, 2011 05:17

8ender: Sep 24, 2003; clown is watching you sleep

NotShadowStar posted:

The asset pipeline is a horribly implemented idea and I hope it either goes away or gets fixed. I keep running into these horror stories repeatedly about it just not working right.

I'm just converting a Rails 2 app over and I've started using the pipeline. Its... okay so far. Lots of magic happening there which makes me nervous. What kind of horror stories are you hearing?

# ? Nov 1, 2011 05:53

NotShadowStar: Sep 20, 2000

The magic just flat out not working, and nobody can figure out WTF like above.

# ? Nov 1, 2011 06:01

8ender: Sep 24, 2003; clown is watching you sleep

Wow day one on the asset pipeline and I really want to disable it. I really really don't want to convert all my css files to use asset_path. Is there a way to disable it specifically for just the images?

# ? Nov 1, 2011 18:30

8ender: Sep 24, 2003; clown is watching you sleep

I bit the bullet and decided to try out the adding .erb to the css files and using the asset tags. Something helpful for anyone doing this:

In Textmate, to find and replace image paths within CSS files with the asset helper:

find: url$\"?(\/?images\/)?(.*)\"?$
replace: url(<%= asset_path("$2") %>)

Does a decent enough job.

So far everything is work but this makes me pretty nervous. The only reason I'm pursuing it is because right now our apps suffer from the exact problem that the pipeline is supposed to solve. We'll see how it goes. For reference we're not using Heroku, but Speedyrails.

# ? Nov 1, 2011 19:17

Molten Llama: Sep 20, 2006

8ender posted:

Wow day one on the asset pipeline and I really want to disable it. I really really don't want to convert all my css files to use asset_path. Is there a way to disable it specifically for just the images?

You shouldn't need to. Unless you're doing something weird, public/ is still used to serve non-JS, non-CSS content.

If you're using regular ol' CSS (and haven't screwed up your routing), /images/ still maps to public/images/.

Disregard, in the same way the current assets pipeline disregards any sense of consistency or logic.

Molten Llama fucked around with this message at 17:10 on Nov 7, 2011

# ? Nov 1, 2011 20:42

NotShadowStar: Sep 20, 2000

8ender posted:

Wow day one on the asset pipeline and I really want to disable it. I really really don't want to convert all my css files to use asset_path. Is there a way to disable it specifically for just the images?

Yeah there it is.

DHH is rather a cock but I hope he has enough class to realize the implementation was total poo poo. :worried:

# ? Nov 2, 2011 00:49

8ender: Sep 24, 2003; clown is watching you sleep

I ended up going for it anyways and converting the css files. Not sure about this but its easy to reverse. Everything seems to be working at the moment.

Goddamn I really shouldn't have waited so long. The jump from 2.3.10 to 3.1.0 is harsh.

# ? Nov 2, 2011 08:02

BrokenDynasty: Dec 25, 2003

How does everyone handle asset discrepancy from production/staging/development for assets not managed in their repository. For example on production there is a CMS where admins (or users) are uploading content, and then when I refresh the development/staging DB with a fresh dump from production, none of the new assets are on the development/staging servers.

I don't really want to copy all the assets over to each environment every time I'm bringing the databases up to snuff, but I also don't want my development/staging environments uploading assets to the same place/bucket as my production app.

Is there some kind of way to cascade requests for assets? Like look for assets at this path, if not found look at this path, globally (as in check the development environment location and if not found then load from production environment location). How do others handle this problem?

# ? Nov 2, 2011 19:42

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

BrokenDynasty posted:

How does everyone handle asset discrepancy from production/staging/development for assets not managed in their repository. For example on production there is a CMS where admins (or users) are uploading content, and then when I refresh the development/staging DB with a fresh dump from production, none of the new assets are on the development/staging servers.

Like, you have code/templates in your app that depend on database/assets in production?

# ? Nov 2, 2011 23:27

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

NotShadowStar posted:

Yeah there it is.

DHH is rather a cock but I hope he has enough class to realize the implementation was total poo poo. :worried:

It's probably not going away; it's much easier to set up than sprockets was (basically zero config), and only requires a bit of a shift in how you work in a new app. It's a bitch getting older/legacy apps configured, yes, but I do like that there's an easy story on doing the full CDN asset thing in production.

# ? Nov 2, 2011 23:29

Melraidin: Oct 11, 2005

I want to restrict access to some rows based on who the current user is. I'm trying to do this by defining a default scope on my model that filters these rows:

code:

unless $current_user.nil?
	self.default_scoping <<	{
		:find => { :conditions => [
				'posts.owner_id NOT IN ( ' + $current_user.hidden_owners.join( ', ' ) + ' ) OR posts.owner_id IS NULL'
			] },
		:create => {}
	}
end

This is Rails 2.3.2. I'm using a global here for the current user object since I haven't found another method of accessing the currently logged in user at this point in the code. I think I'm forced to set the default_scoping member directly since if I were to just use default_scope it would also set the create scope.

Can anyone suggest a way of handling this that isn't quite so awful? I'm worried about relying on a global along with the overall approach.

# ? Nov 4, 2011 16:50

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

Melraidin posted:

I want to restrict access to some rows based on who the current user is. I'm trying to do this by defining a default scope on my model that filters these rows:
code:
unless $current_user.nil?
	self.default_scoping <<	{
		:find => { :conditions => [
				'posts.owner_id NOT IN ( ' + $current_user.hidden_owners.join( ', ' ) + ' ) OR posts.owner_id IS NULL'
			] },
		:create => {}
	}
end
This is Rails 2.3.2. I'm using a global here for the current user object since I haven't found another method of accessing the currently logged in user at this point in the code. I think I'm forced to set the default_scoping member directly since if I were to just use default_scope it would also set the create scope.

Can anyone suggest a way of handling this that isn't quite so awful? I'm worried about relying on a global along with the overall approach.

Figure out how to attach it on the User model instead of the Posts model, or at least in a way you can do current_user.posts.owned.

I never put the current user in a global. It's more common to use a cattr_accessor :current on the User class, but that's still a nasty coupling between the concept of a web session and the business logic.

# ? Nov 4, 2011 17:11

Melraidin: Oct 11, 2005

BonzoESC posted:

Figure out how to attach it on the User model instead of the Posts model, or at least in a way you can do current_user.posts.owned.

I never put the current user in a global. It's more common to use a cattr_accessor :current on the User class, but that's still a nasty coupling between the concept of a web session and the business logic.

I'm very new to Rails so I'm probably missing something but when I tried accessing our current_user object (not the global) in the User model it was inaccessible. I wasn't able to find a way to access this variable that gets instantiated in the application controller from within my model unless I passed it into the model. I wanted to avoid this as I wanted to make use of the default scope to implement the required functionality instead of having to ensure I modified every use of a finder on the User model to ensure they all used a named scope (where I'd be able to pass in the current user).

If I switch to using a class attribute on the User model I'm not sure how I could declare a default scope that would use the current user attribute.

Can you suggest any way I can manage to filter the results of a finder on a model based on the currently logged in user without having to modify every use of a finder on the model throughout the project?

# ? Nov 4, 2011 20:12

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

Melraidin posted:

I'm very new to Rails so I'm probably missing something but when I tried accessing our current_user object (not the global) in the User model it was inaccessible. I wasn't able to find a way to access this variable that gets instantiated in the application controller from within my model unless I passed it into the model. I wanted to avoid this as I wanted to make use of the default scope to implement the required functionality instead of having to ensure I modified every use of a finder on the User model to ensure they all used a named scope (where I'd be able to pass in the current user).

Controller stuff isn't available in models; that's by design, since frequently you'll want to do model things without a controller (maintenance, migrations, background tasks).

Melraidin posted:

If I switch to using a class attribute on the User model I'm not sure how I could declare a default scope that would use the current user attribute.

Can you suggest any way I can manage to filter the results of a finder on a model based on the currently logged in user without having to modify every use of a finder on the model throughout the project?

Move the finder to the current user, and call that one in the controller instead of one that lives on the model's class. If "modify[ing] every use of a finder on the model" is a seemingly-insurmountable task, you're now learning what "technical debt" means.

# ? Nov 4, 2011 20:38

Melraidin: Oct 11, 2005

BonzoESC posted:

Controller stuff isn't available in models; that's by design, since frequently you'll want to do model things without a controller (maintenance, migrations, background tasks).

Move the finder to the current user, and call that one in the controller instead of one that lives on the model's class. If "modify[ing] every use of a finder on the model" is a seemingly-insurmountable task, you're now learning what "technical debt" means.

Oh, there's no doubt I understand technical debt. I've been through one company where I dug them out of a hell of a hole over a few years but now I'm somewhere where they can't even be convinced they have a problem most of the time and, when they do recognize a problem, insist a rewrite's the only way to go. :bang:

Not that I'm necessarily against a rewrite in some cases, but, if you have a large system with poorly hacked together business logic spread literally everywhere in the system how then trying to do a rewrite and match the existing behaviour can be... tough.

Thanks for the help, I'll go back to this thing today and see what can be done to get rid of a little more of the suck.

# ? Nov 7, 2011 15:08

Newbsylberry: Dec 29, 2007; I Swim in drag shorts because I have a SMALL PENIS

So I apologize if someone else has asked this exact questions, or it's been answered. To be honest I just started programming and so everything is still a little hard to decipher.

I am going through the rails 3 tutorial by Michael Hartl, and I can't figure out how to add exceptions to my .autotest file. The issue is that my tests are getting tripped up in app/assets and autotest isn't running at all. Here is what I have in there currently:

Autotest.add_hook :initialize do |autotest|
%w{.git /apps/assets/* .svn .hg .DS_Store ._* vendor tmp log doc}.each do |exception|
autotest.add_exception(exception)
end
end

Let me know if anyone has any ideas.

# ? Nov 8, 2011 18:00

rugbert: Mar 26, 2003; yea, fuck you

I just installed ubuntu 11.04 (down 11.10) and Im having issues with taps when trying to to pull my database from heroku. I keep getting a database connection error, taps is saying it cant find a sequel adapter even tho I have the mysql2 gem installed. Anyone else having this problem?

code:

  Sequel::AdapterNotFound -> LoadError: no such file to load -- mysql2

# ? Nov 8, 2011 18:07

Pardot: Jul 25, 2001

Newbsylberry - Just starting out I wouldn't think you'd need to worry about autotest.

rugbert - Taps :argh:

It's awful awful awful. The usual problem is that it is super slow and often blows up if you have any sort of interesting data. It's always better to use pgbackups to grab a proper dump, even if it's 2 extra steps. I'd recommend using that instead.

I've always had big problems with ruby from apt. Are you using that or a compiled ruby from like rvm?

# ? Nov 8, 2011 19:13

rugbert: Mar 26, 2003; yea, fuck you

Pardot posted:

rugbert - Taps It's awful awful awful. The usual problem is that it is super slow and often blows up if you have any sort of interesting data. It's always better to use pgbackups to grab a proper dump, even if it's 2 extra steps. I'd recommend using that instead.

thanks Ill remember that next time but I just cloned the app (after cloning another one I had forgotten to backup) and it magically started working.

# ? Nov 8, 2011 19:18

A MIRACLE: Sep 17, 2007; All right. It's Saturday night; I have no date, a two-liter bottle of Shasta and my all-Rush mix-tape... Let's rock.

Yeah taps is... kinda bad. Like, it breaks with RefineryCMS is you have any non-default settings set on the server, and then try to db:pull. You have to go into the console, manually delete the settings and then recreate them. Super annoying.

# ? Nov 8, 2011 19:40

Newbsylberry: Dec 29, 2007; I Swim in drag shorts because I have a SMALL PENIS

Pardot posted:

Newbsylberry - Just starting out I wouldn't think you'd need to worry about autotest.

The tutorial has you write different tests, etc. I am thinking of scrapping that aspect of it completely though. It's been incredibly frustrating getting it all to work. The stylesheets issue aside, for a while it was giving me errors incorrectly (like it would say my title wasn't "websitename | title" even though checking the rails server verified that it was), so I had to adjust all the versions I was using in my gem file to get it to pass. Then rake db:migrate wasn't working, because of the version of RSPEC I was using. So it appears that if I'd like to use rake then I won't be able to have accurate testing.

I had an idea for a website a while back and decided to learn how to make it. I think it's a good skill to be able to put on your resume, and I've really enjoyed the process so far. As I get farther and farther along, I realize that to be able to put out a finished product alone I'll need to learn design stuff as well. If I'm going to learn one of the markup and styling languages, which works best with ruby? From what I understand xhtml is the future, would that be best?

one more question: If I skip testing for the tutorial, is there a book/online guide that can help teach me when to use/how to use testing?

Newbsylberry fucked around with this message at 20:02 on Nov 8, 2011

# ? Nov 8, 2011 20:00

Doc Hawkins: Jun 15, 2010; Dashing? But I'm not even moving!

Newbsylberry posted:

The tutorial has you write different tests, etc. I am thinking of scrapping that aspect of it completely though. It's been incredibly frustrating getting it all to work.

That's a shame. I'm not sure what tutorial you're using, so maybe it's a bad one, but I'm going to go ahead and be dogmatic: writing automated tests, preferably before you write anything else, is one of the most important aspects of web development. It's vital, but it shouldn't be that frustrating: I think something's wrong.

quote:

The stylesheets issue aside, for a while it was giving me errors incorrectly (like it would say my title wasn't "websitename | title" even though checking the rails server verified that it was), so I had to adjust all the versions I was using in my gem file to get it to pass.

Okay, it sounds like you had trouble with a view/integration spec. I'll back off my dogmatism for a second: testing views gets you the worst return for your time investment of all the kinds of testing. I still do it, but I try to not be over-specific. For example, I might test that /projects/1 has a title that contains the project name, using a regex and ".should =~" rather than ".should ==".

But that said, I don't know what your gemspec (e: by which of course I mean your gemfile) has to do with page titles. What do you mean by that?

quote:

Then rake db:migrate wasn't working, because of the version of RSPEC I was using. So it appears that if I'd like to use rake then I won't be able to have accurate testing.

Was this the "expected rake 0.9.2" bug, or something else? The rspec-rails gem uses rake extensively, so they definitely do work together.

quote:

I had an idea for a website a while back and decided to learn how to make it. I think it's a good skill to be able to put on your resume, and I've really enjoyed the process so far. As I get farther and farther along, I realize that to be able to put out a finished product alone I'll need to learn design stuff as well. If I'm going to learn one of the markup and styling languages, which works best with ruby? From what I understand xhtml is the future, would that be best?

Well, you'll definitely benefit from understanding html markup, conventions and best practices. But since we're talking about dynamic applications ITT, your real choice is between template languages which generate (x)html. The default Rails templating language is called ERB, and is basically html pages with special dynamic tags that say things like "print the current user here" or "do this next part once for every item in this list" or "only show this next part when the moon is full".

HAML is the new hotness in templates (today, at least), so you could look into that, but I recommend that you defer learning fancy new things that you don't absolutely have to. Just thinking of my own experience, it's possible to overwhelm yourself if you try to use every fancy new great thing. "Learning Rails" is a multifarious and continuous process towards a rapidly moving target, so it's fine to give yourself to a series of bite-size goals.

quote:

one more question: If I skip testing for the tutorial, is there a book/online guide that can help teach me when to use/how to use testing?

I'm halfway through Pragmatic's "Rails Test Prescriptions," and I'm liking it. I also highly recommend their "The RSpec Book" (much of which is about Rails).

Doc Hawkins fucked around with this message at 00:41 on Nov 9, 2011

# ? Nov 8, 2011 21:15

Newbsylberry: Dec 29, 2007; I Swim in drag shorts because I have a SMALL PENIS

Doc Hawkins posted:

Lots of helpful

Luckily I realized I could simply run a spec file with RSPEC and so just because some of my tests aren't giving me accurate results doesn't mean I have to scrap the whole thing.

The gemfile shouldn't have anything to do with it, but apparently following the tutorial leads to errors that have to do with the version of autotest. It doesn't really make sense to me, but I know that changing between version of autotest, spork, and RSPEC means the difference between passing tests, failing tests, or no tests. I think the biggest issue is that I am following a guide, and not fully understanding why it's saying to do what it does.

Maybe it's because the book starts with static pages, but the impression I've gotten is that I'll need to know html and css to deal with the aesthetics of the page. It would be nice if I didn't have to worry about it at all.

# ? Nov 8, 2011 23:40

Doc Hawkins: Jun 15, 2010; Dashing? But I'm not even moving!

Newbsylberry posted:

Luckily I realized I could simply run a spec file with RSPEC and so just because some of my tests aren't giving me accurate results doesn't mean I have to scrap the whole thing.

The gemfile shouldn't have anything to do with it, but apparently following the tutorial leads to errors that have to do with the version of autotest. It doesn't really make sense to me, but I know that changing between version of autotest, spork, and RSPEC means the difference between passing tests, failing tests, or no tests.

Correct me if I'm misunderstanding the situation: You can alter the version requirements in your gemfile, and then run `bundle install`, and no errors are reported, but some tests are provably false. Alternately, you change the version matchers to something else, and run bundle install again, and the tests start working right, but then you can't run any rake tasks.

If that's all accurate, then I would like to see:

Any failed test message of the first case.
The error message of the second case.

This is the paid course at railstutorial.org, right? I am in no way trying to shoo you off, but have you tried to contact the author as well? I hope he'd be able to help you with his own guide...

quote:

I think the biggest issue is that I am following a guide, and not fully understanding why it's saying to do what it does.

...but this does not bode well for that hope. Rails is a big stack, like I said, but a truly good end-to-end introduction shouldn't leave you at sea.

Have you programmed any Ruby separately from Rails?

quote:

Maybe it's because the book starts with static pages, but the impression I've gotten is that I'll need to know html and css to deal with the aesthetics of the page. It would be nice if I didn't have to worry about it at all.

My perspective as a grungy back-end guy is that design is really tough: it's something that some people are born with a talent for, which they hone with years of experience, and when I want my sites to look nice, I need to just hire one of those people.

# ? Nov 9, 2011 00:08

duck monster: Dec 15, 2004

Have any of you people had any luck making rails sane with webfaction. They are a loving good django host but my experience of them with rails has been deranged. Tripping over and smacking my head at the first obstacle of gem not having access to anywhere to write gems, kind of makes rails a bit useless frankly, particularly since what I want is to just get bloody redmine working with a few plugins (yes theres a default redmine installed, no its not any better trying to get plugins installed because its just not recognising gruff which I need.

has anyone else had better experiences, or should I just cancel and go elsewhere?

# ? Nov 10, 2011 13:17

NotShadowStar: Sep 20, 2000

JUST USE HEROKU, gently caress

# ? Nov 10, 2011 16:44

8ender: Sep 24, 2003; clown is watching you sleep

Well the Asset Pipeline sure is a complete bastard. Had to see it for myself. After 10 million commits our app is almost there. Still hanging hard on PDFKit generation if config.assets.compile = true is turned on. Just not working at all if its turned off.

The end product is nice but gently caress the hoops you have to jump through, and the continued poo poo we'll have to do to make sure nothing breaks is just ridiculous.

- There doesn't seem to be a good way to setup different style sheets for different areas aside from loving with manifests inside little empty css files and adding them to config.assets.precompile
- If an asset you're calling is missing in development mode you'll get no warning before a nice hard exception when its not precompiled in production
- config.assets.compile seems to solve this but its slow as hell
- Every stylesheet needed .erb put on it and the image paths rewritten
- CSSEdit doesn't work with the new image paths so I'm doing css with essentially no images in the live preview now
- I need to include gems like a javascript library in my bundle to support precompiling
- Precompiling takes forever on deploy
- Layouts now seem useless as they all just call my main application template now

I mean the finished working pipeline is cool. Chrome is reporting just 2-5kb of loading between pages in our app. Everything else is nicely cached and chrome recognizes and re-downloads the assets when they change. That part is cool. Overall though I'm kind of wishing we hadn't gone to the trouble and waited for the pipeline to be a little more refined.

8ender fucked around with this message at 17:51 on Nov 10, 2011

# ? Nov 10, 2011 17:48

Cock Democracy: Jan 1, 2003; Now that is the finest piece of chilean sea bass I have ever smelled

NotShadowStar posted:

JUST USE HEROKU, gently caress

I really dislike this attitute. Let's face it: Heroku went down for days earlier this year. I realize it wasn't really their fault and they are taking action to prevent it from happening again. I also realize they have some cool features and it's a preference of a lot of people, but it's certainly not the only pain free option. Rails developers shouldn't put all their eggs in one basket when it comes to hosting, right?

That being said, rails is certainly a pain in the rear end on some hosts. I had hytek hosting and had major problems with it, so I got a rackspace cloud server and rolled out my own apache + passenger server. It's a bit more work, sure, but I like having total control and the ability to fix anything myself.

I hear speedyrails is good too. I think a good rule of thumb is to only use rails-centric hosts.

# ? Nov 10, 2011 19:26

NotShadowStar: Sep 20, 2000

On the other hand, I can just do 'git push heroku master' and never have to super sperg about anything.

# ? Nov 10, 2011 19:31

hmm yes: Dec 2, 2000; College Slice

Every host is going to have downtime, the fact that Heroku clearly communicates issues (status.heroku.com is awesome) and makes pushing to production so easy makes it the clear choice for hosting. We're not server administrators at our shop, we're developers and designers, and Heroku lets us keep it that way. Hopefully some clones come out that let us not put all of our eggs in one basket, but until then I will toss them all in with Heroku.

# ? Nov 10, 2011 19:35

soullessshoe: Nov 6, 2011

Anyone know of a good resource for process management in ruby? Going to need to do some forking.

EDIT: Is this a safe/proper way to do it?

code:


children = []
list = [1,2]
list.each do |i|
  children << fork {
    puts i
    100.times {|num| puts "#{i}-#{num}"}
    sleep 1000
  }
end
trap("INT") { 
  children.each {|i| Process.kill("INT", i) }
  exit
}
list = nil
Process.waitall

soullessshoe fucked around with this message at 21:35 on Nov 10, 2011

# ? Nov 10, 2011 20:19

NotShadowStar: Sep 20, 2000

Ruby forks aren't real OS forked processes, they're handled by the Ruby VM. If you're needing to do that for supporting multiple cores, fire up multiple Ruby processes communicating through DRB or something or use NodeJS. It just added true multithreading.

# ? Nov 11, 2011 04:33

Cocoa Crispies: Jul 20, 2001; Vehicular Manslaughter!; Pillbug

NotShadowStar posted:

Ruby forks aren't real OS forked processes, they're handled by the Ruby VM.

code:

puts Process.pid; fork { puts Process.ppid, Process.pid }
71652
71652
71694

# ? Nov 11, 2011 06:35

Adbot: ADBOT LOVES YOU

# ? May 14, 2024 06:37

8ender: Sep 24, 2003; clown is watching you sleep

Cock Democracy posted:

I hear speedyrails is good too. I think a good rule of thumb is to only use rails-centric hosts.

Speedyrails is in fact, fantastic. We've been using them for 2 years now without an issue. They'll setup a server exactly how you want it (unicorn, nginx, passenger, crazy server of the month, whatever) and then from there they'll maintain the server and step in to assist whenever needed.

They setup an app on request (folder structure, vhosts, everything) and send you a capistrano script for deployment. You get whatever access you need to fool around in there and if you gently caress it up they'll step in and fix it. Need an SSL cert installed? Just send it on and its done in a half hour. PDFKit cant find wkhtmltopdf and you suck at linux? Send an email and its solved. Its wonderful.

We've only had one downtime incident with them and it was due to an infrastructure failure in the CANIX data centre. They've since moved to a Peer1 facility. Can't say enough good things about these guys.

For reference we have three production servers and a dev server with them. For capacity they provide a free new relic account and we adjust the servers up and down their tiers as necessary.

# ? Nov 11, 2011 07:07

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Ruby on Rails Love-In

«‹›178 »