RSpec's matchers without all bullshit

I've used to use and like RSpec. The problem was that it always broke with new versions of Rails and internals were too complex to understand. Then appeared Shoulda and removed the pain of writing RUnit tests, so I switched to using it. But I always missed the elegance of RSpec's matchers.

Until Matchy appeared.

The cool thing about it is that you can create custom matchers easily with #custom_matcher() method provided. Nice.

Allowing ssh access to Darcs repositories

I've been using Darcs since early 2006 and I do not see any reason to switch to any other (D)VCS like git.

Today at work we decided to put our new project into Darcs and I needed to setup access for others to my repository. The readonly access is not a problem, but allowing others to push into my repositories was a bit trickier. There is a protocol wrapper called "darcs-server", but I don't like that solution. I wanted to configure Darcs to do get/pull/push with minimal additional components.

The traditional way to do Push is via SSH. You can create a separate account for Darcs VCS or just share your own. Then, you set up access via public/private keys and here we go: you can do all operations. The only thing is that everybody can do anything with your account, which is bad. You need to restrict users to running just a small set of commands sufficient for normal Darcs operation. With Darcs 2 it turns out that you only need to allow running "darcs" with various arguments (as Darcs 2 uses special tunelling to do all work while Darcs 1 used to open tons of connections). After several experiments, here is what you need:

#!/bin/sh

first_word() { echo $1; }

line="$SSH_ORIGINAL_COMMAND"

command=$(first_word $line)
if [ "$command" = "darcs" ]; then
  sh -c "$line"
else
  echo "$command is not allowed"
fi

Then, you add "command='/home/username/bin/darcs-wrapper' " at the beginning of your collegues' public key lines in ~/.ssh/authorized_keys and boom - they are restricted to running only darcs. You can also add no-port-forwarding, no-X11-forwarding, no-agent-forwarding, no-pty to tighten security. And you can set environment variable DARCS_LOGNAME to name of particular user to get proper names in darcs changes logs: add "environment='DARCS_LOGNAME=username'" to your authorized_keys file.

Hope that helps.

UPDATE: after thinking a little, I found a case when this can be workaround to execute any command, e.g. "darcs help; rm -rf *". The problem is with "sh -c". Originally, it was introduced in place of call to "exec" to workaround problem with extra quotation marks that are not removed when parsing line and which cause Darcs to handle arguments improperly. It turns out, that "exec" is really needed there and unquoting should be done manually. Here is the corrected version:

#!/bin/sh

first_word() { echo $1; }
unquoted_words() {
  for word in $@; do
    echo $(echo $word | sed -E "s/^'(.*)'$/\1/; s/^\"(.*)\"$/\1/")
  done
}

line="$SSH_ORIGINAL_COMMAND"

command=$(first_word $line)
if [ "$command" = "darcs" ]; then
  exec $(unquoted_words $line)
else
  echo "$command is not allowed"
fi

UPDATE2: that "sed -E" thing seem to not work on Linux. Linux Sed's option for full regular expressions is called "-r". Probably need to try rewriting regexps to normal Sed's functionality. Later.

UPDATE3: very handy thing in debugging Darcs' SSH is using "--debug --verbose" options to "darcs". Then, you can see what commands are sent to server and how response is interpreted. Then, you can issue those commands manually and see if server's wrapper script outputs errors.

Specific or default templates in Rails

I want to develop admin controller to do simple CRUD operations on not-so-complex models. I want it to be DRY and to allow adding features to all of them easily. This is my second attempt, the first one was way too ugly: it used inheritance from admin's base controller and descendants passed customization parameters to base class; also, have used custom "render" methods to find template for specific controller or, if the specific template is missing, fall back to general template.

And then it broke. Yeah, this could be fixed, but the whole idea seemed bad.

I just finished the second attempt and it looks promising.

First, it includes only one controller and uses routes to determine the actual model to operate on. Every route includes additional parameter that is passed to controller that contains name of model:

map.resources :users, :controller => 'items',
    :requirements => { :item_type => 'User' }
map.resources :regions, :controller => 'items',
    :requirements => { :item_type => 'Region' }

Second, the most troublesome was to make it work like this: I have default views in app/views/items... I wanted to have subdirectories with name of particular model that would contain overrides to default templates:

app/views/items
    index.html.erb
    new.html.erb
    edit.html.erb
    _item.html.erb
    _form.html.erb
app/views/items/users
    _item.html.erb
    _form.html.erb
app/views/items/regions
    _form.html.erb

Regions doesn't need custom 'item' partial because default one (which outputs only item.to_s) fits well.

Obviously, controller's view_paths should be manipulated to search first in concrete subfolders and then in default place. But the problem was that it always tried to prepend controller name and so find it in corresponding subfolder. It's not very nice being forced to create additional directories like this:

app/views/items
app/views/items/users/items
    _item.html.erb

Well, after some digging the way to do it elegantly I found this solution:

1. Redefine controller's controller_path to return nil. After this all templates are searched in app/views/ instead of app/views/items, so it needs to be fixed, so

2. Redefine controller's view_paths to base in app/views/items.

3. Based on current item type prepend corresponding view path.

The result is as follows:

class ItemsController < ApplicationController
  def self.controller_path
    nil
  end
  self.view_paths = [ RAILS_ROOT + '/app/views/items' ]

  before_filter :setup_model

  # ... action code

  private

  def setup_model
    @model  = params[:item_type].constantize
    prepend_view_path(RAILS_ROOT + '/app/views/items/' + params[:item_type].underscore)
  end
end

Happy coding!

Back to Rails

Lately I wasn't doing any Rails development because I was busy re-learning C/C++ "the right way". Turns out it's not that painful as they say.

Then I had to do some Rails coding using latest and greatest Rails 2.2 and I find that that the simple things are broken (or somewhat obscure).

Case 1.

I was developing some general resource controller and wanted to spec out a redirect after #create action. My controller should not be bound to any specific resource url (as it is planned to bind many different resources with that controller), so I decided to make use of Rails' url rewriting. Spec was like this:

it "should redirect to #index after item creation" do

  @model.stubs(:create!)


  post :create

  response.should redirect_to(:action => :index)

end

And this spec failed because of MethodNotAllowed: Only get and post requests are allowed.

WTF? It's the very basic functionality. Why does it fail to work ?

After half a day of investigation I found out that this is a bug in rspec-rails package and redirect_to matcher fails to handle hash-like url specifier.

Case 2.

I wanted to handle AR::RecordInvalid exceptions with one handler to DRY up controller code. I wrote corresponding "rescue_from" call in controller and wrote specs (in fact, the reverse order: specs then code). And those specs failed too.

The other several hours of investigation revealed hidden option: RSpec-rails overrides default Rails' error handling (which is based on ActiveSupport's #rescue_from feature) with pass-through exception handler, and you have to write following code to make YOUR exception handlers work:

before do

  controller.use_rails_error_handling!

end

Yeah, Rails is nice, but upgrading to a new version is a pain.

Sphinx FAIL

On my recent Rails project I decided to try Sphinx search engine. Before that I used Ferret and then - SOLR. I abandoned Ferret because it's instability and lack of tools to track problems (like Luke - index browser for Lucene).

So, Sphinx. I have found two plugins for Rails - acts_as_sphinx and Unltrasphinx. Ok, I've heard in mailing lists that Ultrasphinx is better (and Sphinx recipe in Advanced Rails Recipes book also involves it) so I decided to use it.

First I have to install Sphinx from sources because version in MacPorts (although, latest released version) is too old and Ultrasphinx requiresa newer one (at that point - release candidate 2 of next revision, which is 0.9.8rc2 vs 0.9.7). Then I had to do various dances around Sphinx to compile it on my Mac OS X (which are described in my previous post.).

And then it began:

1. It was needed that new data (e.g. new articles) could be searched with Sphinx right after it was added. And then I find out, that it is not encouraged to do such updates often, you'd better run full updates once a day... Wtf? I have heard that there are something called "deltas", but as I found out from plugin, it doesn't install any hooks on models, so I assume that deltas need to be built periodically (which is also unacceptable).
After consulting with other people who have used Sphinx I found out that:
1) they don't use plugins and do all communication with some low level library (Riddle, as far as I remember) manually.
2) they install their own hooks on models and call indexer manually to reindex their models. I have tried to to install after_save hook but it is run BEFORE transaction is commited and indexer can't see inserted/updated data, so I don't know how this can be accomplished easily.

2. Sphinx configuration is one big scary thing. Although Ultrasphinx managed to build it for me, I needed to do some tweaks to it and then next time I needed to build configuration for another model my tweaks were lost.

3. Model on which I have used Ultrasphinx (called ::is_indexed) failed to load by automatic dependency loader. After several hours of tracking this problem I stuck "require 'my_model'" into environment.rb (which helped) and cursed that Sphinx and all it's plugins.

So, for me Sphinx definitely FAIL.

PS
I have tried SOLR and it worked like charm:
1) almost no configuration.
SOLR configuration consists of type definitions (which describe how value of that type should be analyzed), field definitions and dynamic field definitions (acts_as_solr rails plugin uses only dynamic fields). Acts_as_solr comes with default SOLR configuration which contains commonly used types and dynamic field definitions for them.
2) no compilation needed. Acts_as_solr comes with SOLR JAR files, so you just need proper version of Java runtime installed.
3) works like charm. Everything you would expect from full-text search engine.