active_column 0.0.2 → 0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +1 -0
- data/.yardopts +1 -0
- data/Gemfile.lock +120 -17
- data/README.md +27 -143
- data/active_column.gemspec +12 -6
- data/docs/Create.md +101 -0
- data/docs/Migrate.md +100 -0
- data/docs/Query.md +43 -0
- data/lib/active_column.rb +22 -3
- data/lib/active_column/base.rb +32 -46
- data/lib/active_column/errors.rb +6 -0
- data/lib/active_column/generators/migration_generator.rb +31 -0
- data/lib/active_column/generators/templates/migration.rb.erb +11 -0
- data/lib/active_column/key_config.rb +16 -0
- data/lib/active_column/migration.rb +269 -0
- data/lib/active_column/tasks/column_family.rb +64 -0
- data/lib/active_column/tasks/keyspace.rb +59 -0
- data/lib/active_column/tasks/ks.rb +76 -0
- data/lib/active_column/version.rb +1 -1
- data/spec/active_column/base_crud_spec.rb +1 -1
- data/spec/active_column/base_finders_spec.rb +6 -6
- data/spec/active_column/migrator_spec.rb +150 -0
- data/spec/active_column/tasks/column_family_spec.rb +34 -0
- data/spec/active_column/tasks/keyspace_spec.rb +38 -0
- data/spec/spec_helper.rb +24 -4
- data/spec/support/aggregating_tweet.rb +3 -1
- data/spec/support/migrate/migrator_spec/1_migration1.rb +11 -0
- data/spec/support/migrate/migrator_spec/2_migration2.rb +11 -0
- data/spec/support/migrate/migrator_spec/3_migration3.rb +11 -0
- data/spec/support/migrate/migrator_spec/4_migration4.rb +11 -0
- data/spec/support/tweet.rb +4 -2
- data/spec/support/tweet_dm.rb +6 -4
- metadata +103 -11
data/.yardopts
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
--no-private --protected - docs/*.md
|
data/Gemfile.lock
CHANGED
@@ -1,38 +1,141 @@
|
|
1
1
|
PATH
|
2
2
|
remote: .
|
3
3
|
specs:
|
4
|
-
active_column (0.0.
|
5
|
-
|
4
|
+
active_column (0.0.3)
|
5
|
+
activesupport
|
6
|
+
cassandra (>= 0.9)
|
7
|
+
rake
|
6
8
|
|
7
9
|
GEM
|
8
10
|
remote: http://rubygems.org/
|
9
11
|
specs:
|
10
|
-
|
12
|
+
ParseTree (3.0.6)
|
13
|
+
RubyInline (>= 3.7.0)
|
14
|
+
sexp_processor (>= 3.0.0)
|
15
|
+
RubyInline (3.8.6)
|
16
|
+
ZenTest (~> 4.3)
|
17
|
+
ZenTest (4.4.2)
|
18
|
+
abstract (1.0.0)
|
19
|
+
actionmailer (3.0.3)
|
20
|
+
actionpack (= 3.0.3)
|
21
|
+
mail (~> 2.2.9)
|
22
|
+
actionpack (3.0.3)
|
23
|
+
activemodel (= 3.0.3)
|
24
|
+
activesupport (= 3.0.3)
|
25
|
+
builder (~> 2.1.2)
|
26
|
+
erubis (~> 2.6.6)
|
27
|
+
i18n (~> 0.4)
|
28
|
+
rack (~> 1.2.1)
|
29
|
+
rack-mount (~> 0.6.13)
|
30
|
+
rack-test (~> 0.5.6)
|
31
|
+
tzinfo (~> 0.3.23)
|
32
|
+
activemodel (3.0.3)
|
33
|
+
activesupport (= 3.0.3)
|
34
|
+
builder (~> 2.1.2)
|
35
|
+
i18n (~> 0.4)
|
36
|
+
activerecord (3.0.3)
|
37
|
+
activemodel (= 3.0.3)
|
38
|
+
activesupport (= 3.0.3)
|
39
|
+
arel (~> 2.0.2)
|
40
|
+
tzinfo (~> 0.3.23)
|
41
|
+
activeresource (3.0.3)
|
42
|
+
activemodel (= 3.0.3)
|
43
|
+
activesupport (= 3.0.3)
|
44
|
+
activesupport (3.0.3)
|
45
|
+
arel (2.0.6)
|
46
|
+
bluecloth (2.0.9)
|
47
|
+
builder (2.1.2)
|
48
|
+
cassandra (0.9.0)
|
11
49
|
json
|
12
50
|
rake
|
13
51
|
simple_uuid (>= 0.1.0)
|
14
|
-
thrift_client (>= 0.
|
52
|
+
thrift_client (>= 0.6.0)
|
15
53
|
diff-lcs (1.1.2)
|
54
|
+
erubis (2.6.6)
|
55
|
+
abstract (>= 1.0.0)
|
56
|
+
file-tail (1.0.5)
|
57
|
+
spruz (>= 0.1.0)
|
58
|
+
i18n (0.5.0)
|
16
59
|
json (1.4.6)
|
60
|
+
mail (2.2.12)
|
61
|
+
activesupport (>= 2.3.6)
|
62
|
+
i18n (>= 0.4.0)
|
63
|
+
mime-types (~> 1.16)
|
64
|
+
treetop (~> 1.4.8)
|
65
|
+
mime-types (1.16)
|
66
|
+
polyglot (0.3.1)
|
67
|
+
predicated (0.2.2)
|
68
|
+
rack (1.2.1)
|
69
|
+
rack-mount (0.6.13)
|
70
|
+
rack (>= 1.0.0)
|
71
|
+
rack-test (0.5.6)
|
72
|
+
rack (>= 1.0)
|
73
|
+
rails (3.0.3)
|
74
|
+
actionmailer (= 3.0.3)
|
75
|
+
actionpack (= 3.0.3)
|
76
|
+
activerecord (= 3.0.3)
|
77
|
+
activeresource (= 3.0.3)
|
78
|
+
activesupport (= 3.0.3)
|
79
|
+
bundler (~> 1.0)
|
80
|
+
railties (= 3.0.3)
|
81
|
+
railties (3.0.3)
|
82
|
+
actionpack (= 3.0.3)
|
83
|
+
activesupport (= 3.0.3)
|
84
|
+
rake (>= 0.8.7)
|
85
|
+
thor (~> 0.14.4)
|
17
86
|
rake (0.8.7)
|
18
|
-
rspec (2.
|
19
|
-
rspec-core (~> 2.
|
20
|
-
rspec-expectations (~> 2.
|
21
|
-
rspec-mocks (~> 2.
|
22
|
-
rspec-core (2.
|
23
|
-
rspec-expectations (2.
|
87
|
+
rspec (2.3.0)
|
88
|
+
rspec-core (~> 2.3.0)
|
89
|
+
rspec-expectations (~> 2.3.0)
|
90
|
+
rspec-mocks (~> 2.3.0)
|
91
|
+
rspec-core (2.3.1)
|
92
|
+
rspec-expectations (2.3.0)
|
24
93
|
diff-lcs (~> 1.1.2)
|
25
|
-
rspec-mocks (2.
|
94
|
+
rspec-mocks (2.3.0)
|
95
|
+
rspec-rails (2.3.1)
|
96
|
+
actionpack (~> 3.0)
|
97
|
+
activesupport (~> 3.0)
|
98
|
+
railties (~> 3.0)
|
99
|
+
rspec (~> 2.3.0)
|
100
|
+
ruby2ruby (1.2.5)
|
101
|
+
ruby_parser (~> 2.0)
|
102
|
+
sexp_processor (~> 3.0)
|
103
|
+
ruby_parser (2.0.5)
|
104
|
+
sexp_processor (~> 3.0)
|
105
|
+
sexp_processor (3.0.5)
|
26
106
|
simple_uuid (0.1.1)
|
27
|
-
|
28
|
-
|
29
|
-
|
107
|
+
sourcify (0.4.0)
|
108
|
+
ruby2ruby (>= 1.2.5)
|
109
|
+
sexp_processor (>= 3.0.5)
|
110
|
+
spruz (0.2.2)
|
111
|
+
thor (0.14.6)
|
112
|
+
thrift (0.5.0)
|
113
|
+
thrift_client (0.6.0)
|
114
|
+
thrift (~> 0.5.0)
|
115
|
+
treetop (1.4.9)
|
116
|
+
polyglot (>= 0.3.1)
|
117
|
+
tzinfo (0.3.23)
|
118
|
+
wrong (0.5.0)
|
119
|
+
ParseTree (~> 3.0)
|
120
|
+
diff-lcs (~> 1.1.2)
|
121
|
+
file-tail (~> 1.0)
|
122
|
+
predicated (>= 0.2.2)
|
123
|
+
ruby2ruby (~> 1.2)
|
124
|
+
ruby_parser (~> 2.0.4)
|
125
|
+
sexp_processor (~> 3.0)
|
126
|
+
sourcify (>= 0.3.0)
|
127
|
+
yard (0.6.4)
|
30
128
|
|
31
129
|
PLATFORMS
|
32
130
|
ruby
|
33
131
|
|
34
132
|
DEPENDENCIES
|
35
133
|
active_column!
|
36
|
-
|
37
|
-
|
38
|
-
|
134
|
+
activesupport
|
135
|
+
bluecloth
|
136
|
+
cassandra (>= 0.9)
|
137
|
+
rails (>= 3.0)
|
138
|
+
rake
|
139
|
+
rspec-rails
|
140
|
+
wrong
|
141
|
+
yard
|
data/README.md
CHANGED
@@ -1,8 +1,21 @@
|
|
1
|
+
**IMPORTANT**: If you are reading this on the main ActiveColumn page on github, please go to
|
2
|
+
[the actual README page](./active_column/blob/master/README.md) so that links bring you to the right place.
|
3
|
+
|
1
4
|
# ActiveColumn
|
2
5
|
|
3
|
-
ActiveColumn is a framework for
|
4
|
-
|
5
|
-
|
6
|
+
ActiveColumn is a framework for working with data in Cassandra. It currently includes two features:
|
7
|
+
|
8
|
+
- Database migrations
|
9
|
+
- "Time line" model data management
|
10
|
+
|
11
|
+
Data migrations are very similar to those in ActiveRecord, and are documented in [Migrate](./docs/Migrate.md).
|
12
|
+
|
13
|
+
Time line data management is loosely based on concepts in ActiveRecord, but is adapted to saving data in which rows in
|
14
|
+
Cassandra grow indefinitely over time, such as in the oft-used Twitter example for Cassandra. This usage is documented
|
15
|
+
in:
|
16
|
+
|
17
|
+
- [Create](./docs/Create.md) - how to create data
|
18
|
+
- [Query](./docs/Query.md) - how to find data
|
6
19
|
|
7
20
|
## Installation
|
8
21
|
|
@@ -20,20 +33,15 @@ bundle install
|
|
20
33
|
|
21
34
|
### Configuration
|
22
35
|
|
23
|
-
ActiveColumn requires the [cassandra gem](https://github.com/fauna/cassandra)
|
24
|
-
|
25
|
-
|
26
|
-
<pre>
|
27
|
-
ActiveColumn.connection = Cassandra.new('my_keyspace', '127.0.0.1:9160')
|
28
|
-
</pre>
|
36
|
+
ActiveColumn requires Cassandra 0.7 or above, as we as the [cassandra gem](https://github.com/fauna/cassandra),
|
37
|
+
version 0.9 or above.
|
29
38
|
|
30
|
-
|
31
|
-
|
39
|
+
Data migrations in ActiveColumn are used within a Rails project, and are driven off of a configuration file,
|
40
|
+
config/cassandra.yml. It should look something like this:
|
32
41
|
|
33
|
-
|
42
|
+
_config/cassandra.yml_
|
34
43
|
<pre>
|
35
44
|
test:
|
36
|
-
home: ":"
|
37
45
|
servers: "127.0.0.1:9160"
|
38
46
|
keyspace: "myapp_test"
|
39
47
|
thrift:
|
@@ -41,7 +49,6 @@ test:
|
|
41
49
|
retries: 2
|
42
50
|
|
43
51
|
development:
|
44
|
-
home: ":"
|
45
52
|
servers: "127.0.0.1:9160"
|
46
53
|
keyspace: "myapp_development"
|
47
54
|
thrift:
|
@@ -49,7 +56,11 @@ development:
|
|
49
56
|
retries: 2
|
50
57
|
</pre>
|
51
58
|
|
52
|
-
|
59
|
+
In order to get time line modeling support, you must provide ActiveColumn with an instance of a Cassandra object.
|
60
|
+
Since you have your cassandra.yml from above, you can do this very simply like this:
|
61
|
+
|
62
|
+
|
63
|
+
_config/initializers/cassandra.rb_
|
53
64
|
<pre>
|
54
65
|
config = YAML.load_file(Rails.root.join("config", "cassandra.yml"))[Rails.env]
|
55
66
|
$cassandra = Cassandra.new(config['keyspace'],
|
@@ -63,131 +74,4 @@ As you can see, I create a global $cassandra variable, which I use in my tests t
|
|
63
74
|
|
64
75
|
One other thing to note is that you obviously must have Cassandra installed and running! Please take a look at the
|
65
76
|
[mama_cass gem](https://github.com/carbonfive/mama_cass) for a quick way to get up and running with Cassandra for
|
66
|
-
development and testing.
|
67
|
-
|
68
|
-
### Saving data
|
69
|
-
|
70
|
-
To make a model in to an ActiveColumn model, just extend ActiveColumn::Base, and provide two pieces of information:
|
71
|
-
|
72
|
-
- Column Family
|
73
|
-
- Function(s) to generate keys for your rows of data
|
74
|
-
|
75
|
-
The most basic form of using ActiveColumn looks like this:
|
76
|
-
<pre>
|
77
|
-
class Tweet < ActiveColumn::Base
|
78
|
-
column_family :tweets
|
79
|
-
key :user_id
|
80
|
-
end
|
81
|
-
</pre>
|
82
|
-
|
83
|
-
Then in your app you can create and save a tweet like this:
|
84
|
-
<pre>
|
85
|
-
tweet = Tweet.new( :user_id => 'mwynholds', :message => "I'm going for a bike ride" )
|
86
|
-
tweet.save
|
87
|
-
</pre>
|
88
|
-
|
89
|
-
When you run #save, ActiveColumn saves a new column in the "tweets" column family in the row with key "mwynholds". The
|
90
|
-
content of the row is the Tweet instance JSON-encoded.
|
91
|
-
|
92
|
-
*Key Generator Functions*
|
93
|
-
|
94
|
-
This is great, but quite often you want to save the content in multiple rows for the sake of speedy lookups. This is
|
95
|
-
basically de-normalizing data, and is extremely common in Cassandra data. ActiveColumn lets you do this quite easily
|
96
|
-
by telling it the name of a function to use to generate the keys during a save. It works like this:
|
97
|
-
|
98
|
-
<pre>
|
99
|
-
class Tweet < ActiveColumn::Base
|
100
|
-
column_family :tweets
|
101
|
-
key :user_id, :values => :generate_user_keys
|
102
|
-
|
103
|
-
def generate_user_keys
|
104
|
-
[ attributes[:user_id], 'all']
|
105
|
-
end
|
106
|
-
end
|
107
|
-
</pre>
|
108
|
-
|
109
|
-
The code to save the tweet is the same as the previous example, but now it saves the tweet in both the "mwynholds" row
|
110
|
-
and the "all" row. This way, you can pull out the last 20 of all tweets quite easily (assuming you needed to do this
|
111
|
-
in your app).
|
112
|
-
|
113
|
-
*Compound Keys*
|
114
|
-
|
115
|
-
In some cases you may want to have your rows keyed by multiple values. ActiveColumn supports compound keys,
|
116
|
-
and looks like this:
|
117
|
-
|
118
|
-
<pre>
|
119
|
-
class TweetDM < ActiveColumn::Base
|
120
|
-
column_family :tweet_dms
|
121
|
-
key :user_id, :values => :generate_user_keys
|
122
|
-
key :recipient_id, :values => :recipient_ids
|
123
|
-
|
124
|
-
def generate_user_keys
|
125
|
-
[ attributes[:user_id], 'all ]
|
126
|
-
end
|
127
|
-
end
|
128
|
-
</pre>
|
129
|
-
|
130
|
-
Now, when you create a new TweetDM, it might look like this:
|
131
|
-
|
132
|
-
<pre>
|
133
|
-
dm = TweetDM.new( :user_id => 'mwynholds', :recipient_ids => [ 'fsinatra', 'dmartin' ], :message => "Let's go to Vegas" )
|
134
|
-
</pre>
|
135
|
-
|
136
|
-
This tweet direct message will saved to four different rows in the "tweet_dms" column family, under these keys:
|
137
|
-
|
138
|
-
- mwynholds:fsinatra
|
139
|
-
- mwynholds:dmartin
|
140
|
-
- all:fsinatra
|
141
|
-
- all:dmartin
|
142
|
-
|
143
|
-
Now my app can pretty easily figure find all DMs I sent to Old Blue Eyes, or to Dino, and it can also easily find all
|
144
|
-
DMs sent from *anyone* to Frank or Dino.
|
145
|
-
|
146
|
-
One thing to note about the TweetDM class above is that the "keys" configuration at the top looks a little uglier than
|
147
|
-
before. If you have a compound key and any of the keys have custom key generators, you need to pass in an array of
|
148
|
-
single-element hashes. This is in place to support Ruby 1.8, which does not have ordered hashes. Making sure the keys
|
149
|
-
are ordered is necessary to keep the compounds keys canonical (ie: deterministic).
|
150
|
-
|
151
|
-
### Finding data
|
152
|
-
|
153
|
-
Ok, congratulations - now you have a bunch of fantastic data in Cassandra. How do you get it out? ActiveColumn can
|
154
|
-
help you here too.
|
155
|
-
|
156
|
-
Here is how you look up data that have a simple key:
|
157
|
-
|
158
|
-
<pre>
|
159
|
-
tweets = Tweet.find( 'mwynholds', :reversed => true, :count => 3 )
|
160
|
-
</pre>
|
161
|
-
|
162
|
-
This code will find the last 10 tweets for the 'mwynholds' user in reverse order. It comes back as a hash of arrays,
|
163
|
-
and would looks like this if represented in JSON:
|
164
|
-
|
165
|
-
<pre>
|
166
|
-
{
|
167
|
-
'mwynholds': [ { 'user_id': 'mwynholds', 'message': 'I\'m going to bed now' },
|
168
|
-
{ 'user_id': 'mwynholds', 'message': 'It\'s lunch time' },
|
169
|
-
{ 'user_id': 'mwynholds', 'message': 'Just woke up' } ]
|
170
|
-
}
|
171
|
-
</pre>
|
172
|
-
|
173
|
-
Here are some other examples and their return values:
|
174
|
-
|
175
|
-
<pre>
|
176
|
-
Tweet.find( [ 'mwynholds', 'all' ], :count => 2 )
|
177
|
-
|
178
|
-
{
|
179
|
-
'mwynholds': [ { 'user_id': 'mwynholds', 'message': 'Good morning' },
|
180
|
-
{ 'user_id': 'mwynholds', 'message': 'Good afternoon' } ],
|
181
|
-
'all': [ { 'user_id': 'mwynholds', 'message': 'Good morning' },
|
182
|
-
'user_id': 'bmurray', 'message': 'Who ya gonna call!' } ]
|
183
|
-
}
|
184
|
-
</pre>
|
185
|
-
|
186
|
-
<pre>
|
187
|
-
Tweet.find( { 'user_id' => 'all', 'recipient_id' => [ 'fsinatra', 'dmartin' ] }, :reversed => true, :count => 1 )
|
188
|
-
|
189
|
-
{
|
190
|
-
'all:fsinatra' => [ { 'user_id': 'mwynholds', 'recipient_ids' => [ 'fsinatra', 'dmartin' ], 'message' => 'Here we come Vegas!' } ],
|
191
|
-
'all:dmartin' => [ { 'user_id': 'fsinatra', 'recipient_ids' => [ 'dmartin' ], 'message' => 'Vegas was fun' } ]
|
192
|
-
}
|
193
|
-
</pre>
|
77
|
+
development and testing.
|
data/active_column.gemspec
CHANGED
@@ -9,8 +9,8 @@ Gem::Specification.new do |s|
|
|
9
9
|
s.authors = ["Michael Wynholds"]
|
10
10
|
s.email = ["mike@wynholds.com"]
|
11
11
|
s.homepage = "https://github.com/carbonfive/active_column"
|
12
|
-
s.summary = %q{Provides time line support for Cassandra}
|
13
|
-
s.description = %q{Provides time line support for Cassandra}
|
12
|
+
s.summary = %q{Provides time line support and database migrations for Cassandra}
|
13
|
+
s.description = %q{Provides time line support and database migrations for Cassandra}
|
14
14
|
|
15
15
|
s.rubyforge_project = "active_column"
|
16
16
|
|
@@ -18,9 +18,15 @@ Gem::Specification.new do |s|
|
|
18
18
|
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
|
19
19
|
s.executables = `git ls-files -- bin/*`.split("\n").map{ |f| File.basename(f) }
|
20
20
|
s.require_paths = ["lib"]
|
21
|
+
s.has_rdoc = true
|
21
22
|
|
22
|
-
s.add_dependency '
|
23
|
-
s.add_dependency '
|
24
|
-
|
25
|
-
|
23
|
+
s.add_dependency 'cassandra', '>= 0.9'
|
24
|
+
s.add_dependency 'activesupport'
|
25
|
+
s.add_dependency 'rake'
|
26
|
+
|
27
|
+
s.add_development_dependency 'rails', '>= 3.0'
|
28
|
+
s.add_development_dependency 'rspec-rails'
|
29
|
+
s.add_development_dependency 'wrong'
|
30
|
+
s.add_development_dependency 'yard'
|
31
|
+
s.add_development_dependency 'bluecloth'
|
26
32
|
end
|
data/docs/Create.md
ADDED
@@ -0,0 +1,101 @@
|
|
1
|
+
### Saving data
|
2
|
+
|
3
|
+
To make a model in to an ActiveColumn model, just extend ActiveColumn::Base, and provide two pieces of information:
|
4
|
+
|
5
|
+
- Column Family (optional)
|
6
|
+
- Function(s) to generate keys for your rows of data
|
7
|
+
|
8
|
+
If you do not specify a column family, it will default to the "tabelized" class name, just like ActiveRecord.
|
9
|
+
Example: Tweet --> tweets
|
10
|
+
Example: TweetDM --> tweet_dms
|
11
|
+
|
12
|
+
The most basic form of using ActiveColumn looks like this:
|
13
|
+
<pre>
|
14
|
+
class Tweet < ActiveColumn::Base
|
15
|
+
key :user_id
|
16
|
+
attr_accessor :user_id, :message
|
17
|
+
end
|
18
|
+
</pre>
|
19
|
+
|
20
|
+
Note that you can also use ActiveColumn as a mix-in, like this:
|
21
|
+
<pre>
|
22
|
+
class Tweet
|
23
|
+
include ActiveColumn
|
24
|
+
|
25
|
+
key :user_id
|
26
|
+
attr_accessor :user_id, :message
|
27
|
+
end
|
28
|
+
</pre>
|
29
|
+
|
30
|
+
Then in your app you can create and save a tweet like this:
|
31
|
+
<pre>
|
32
|
+
tweet = Tweet.new( :user_id => 'mwynholds', :message => "I'm going for a bike ride" )
|
33
|
+
tweet.save
|
34
|
+
</pre>
|
35
|
+
|
36
|
+
When you run #save, ActiveColumn saves a new column in the "tweets" column family in the row with key "mwynholds". The
|
37
|
+
content of the row is the Tweet instance JSON-encoded.
|
38
|
+
|
39
|
+
*Key Generator Functions*
|
40
|
+
|
41
|
+
This is great, but quite often you want to save the content in multiple rows for the sake of speedy lookups. This is
|
42
|
+
basically de-normalizing data, and is extremely common in Cassandra data. ActiveColumn lets you do this quite easily
|
43
|
+
by telling it the name of a function to use to generate the keys during a save. It works like this:
|
44
|
+
|
45
|
+
<pre>
|
46
|
+
class Tweet
|
47
|
+
include ActiveColumn
|
48
|
+
|
49
|
+
key :user_id, :values => :generate_user_keys
|
50
|
+
attr_accessor :user_id, :message
|
51
|
+
|
52
|
+
def generate_user_keys
|
53
|
+
[ user_id, 'all']
|
54
|
+
end
|
55
|
+
end
|
56
|
+
</pre>
|
57
|
+
|
58
|
+
The code to save the tweet is the same as the previous example, but now it saves the tweet in both the "mwynholds" row
|
59
|
+
and the "all" row. This way, you can pull out the last 20 of all tweets quite easily (assuming you needed to do this
|
60
|
+
in your app).
|
61
|
+
|
62
|
+
*Compound Keys*
|
63
|
+
|
64
|
+
In some cases you may want to have your rows keyed by multiple values. ActiveColumn supports compound keys,
|
65
|
+
and looks like this:
|
66
|
+
|
67
|
+
<pre>
|
68
|
+
class TweetDM
|
69
|
+
include ActiveColumn
|
70
|
+
|
71
|
+
column_family :tweet_dms
|
72
|
+
key :user_id, :values => :generate_user_keys
|
73
|
+
key :recipient_id, :values => :recipient_ids
|
74
|
+
attr_accessor :user_id, :recipient_ids, :message
|
75
|
+
|
76
|
+
def generate_user_keys
|
77
|
+
[ user_id, 'all ]
|
78
|
+
end
|
79
|
+
end
|
80
|
+
</pre>
|
81
|
+
|
82
|
+
Now, when you create a new TweetDM, it might look like this:
|
83
|
+
|
84
|
+
<pre>
|
85
|
+
dm = TweetDM.new( :user_id => 'mwynholds', :recipient_ids => [ 'fsinatra', 'dmartin' ], :message => "Let's go to Vegas" )
|
86
|
+
</pre>
|
87
|
+
|
88
|
+
This tweet direct message will saved to four different rows in the "tweet_dms" column family, under these keys:
|
89
|
+
|
90
|
+
- mwynholds:fsinatra
|
91
|
+
- mwynholds:dmartin
|
92
|
+
- all:fsinatra
|
93
|
+
- all:dmartin
|
94
|
+
|
95
|
+
Now my app can pretty easily figure find all DMs I sent to Old Blue Eyes, or to Dino, and it can also easily find all
|
96
|
+
DMs sent from *anyone* to Frank or Dino.
|
97
|
+
|
98
|
+
One thing to note about the TweetDM class above is that the "keys" configuration at the top looks a little uglier than
|
99
|
+
before. If you have a compound key and any of the keys have custom key generators, you need to pass in an array of
|
100
|
+
single-element hashes. This is in place to support Ruby 1.8, which does not have ordered hashes. Making sure the keys
|
101
|
+
are ordered is necessary to keep the compounds keys canonical (ie: deterministic).
|