Robby on Rails

Aah, pretty neat!

Awesome plugin, just what I was looking for ;)

A cursory glance at the code shows me that this will fail to handle collections properly. Setting a collection to an explicit list will cause a delete_all to be run. Deletes don’t have callbacks and therefor you will wind up hitting the slave. I have filed a bug.

Why wouldn’t you want this turned on for all your models? Is it possible to use delegates_connection_to for all of them?

Can you explain the “&” and ”<<” stuff? I think I’ve seen this before but have never been clear as to how it works.

Can you explain the “&” and ”<<” stuff? I think I’ve seen this before but have never been clear as to how it works.

I think you should use replication features provided by database rather than create one on your own. It will be safer and provide higher performance.

Not everything can be done by standard replication features. Thanks for the tip Robby!

Does this accomplish anything more than http://drnicwilliams.com/2007/04/12/magic-multi-connections-a-facility-in-rails-to-talk-to-more-than-one-database-at-a-time/ ?

@Frank

doing something like login: &login adaptor: mysql etc: etc

then development: database: hat_development <<: *login

So we use & to store a bunch of yaml, and <<: * to put it back in place. I’m not sure the actual code behind it - whether it’s an actual block or if it’s just yaml - but either way, that’s what it’s for.

Think it originally showed up on Redhanded?

Congratulations to writing your first blog post without “PLANET ARGON” in all caps in the body. That line always turns me off enough that I never read beyond it. This time I could read the whole post, and it was interesting. Thanks!

@Jeff: I don’t think it does any more than Dr. Nic’s magic multi connections, but sounds like it has more issues. However, if you are really trying to handle replicated databases seriously I quite like the approach taken in MySQL Replication Adapter project (http://rubyforge.org/projects/mysql-replicate/). It feels much more complete than doing noob AR method extensions to attempt something much larger.

Superb plugin! I had to change mysqldump to mysqldump5 and added the support to only dump the structure of the database “mysql-style” not schema.rb:

desc "Dump only them schema and to an SQL file (/db/schema_YYYY_MM_DD.sql)" 
task :structure => :environment do
  current_date = Time.now.strftime("%Y_%m_%d")
  archive = "#{RAILS_ROOT}/db/schema_#{current_date}.sql" 
  database, user, password = retrieve_db_info
end

cmd = "/usr/bin/env mysqldump --opt --skip-add-locks --no-data -u #{user} " 
puts cmd + "... [password filtered]" 
cmd += " -p'#{password}' " unless password.nil?
cmd += " #{database} > #{archive}" 
result = system(cmd)

Is this sufficient to connect two databases in one rails app? Could you please explain ? I am experiencing the same issue so need your help?

Thanks Saket

Is this sufficient to connect two databases in one rails app? Could you please explain ? I am experiencing the same issue so need your help?

Thanks Saket

nice share

Robby thanks for a great plugin!

# app/models/animal.rb
class Animal < ActiveRecord::Base
  delegates_connection_to :master_database, :on => [:create, :save, :destroy]
end

# app/models/animal.rb
class Animal < MasterDatabase
end

Cool stuff. How does this affect testing with multiple databases?