Minor improvement to index_exists? using correct es api call #51
Commits on Mar 25, 2013
-
Allow multi_get to take a hash
This patch lets you pass arbitrary multi_get queries to elasticsearch. Before this patch, there was no way to select the fields to return.
-
-
-
-
-
-
-
set default index and type from server url
This is helpful on Heroku where server urls include the index. ElasticSearch.new("http://127.0.0.1:9200/index/type" sets default_index to "index" and default_type to "type". You can still change them with the default scope accessors and by passing :index and :type options. Fixes #31 -
don't require a specific bundler
Bundler 1.1 is the business.
-
-
update README to show new url instantiation
README needs some love! It kinda sucks.
-
New in this version: * Compatibility with Heroku ELASTICSEARCH_URL
-
-
-
Will need to switch to http and json abstraction libraries first.
-
-
Fix error in README.md (otherwise: "URI::InvalidURIError: bad URI(is …
…not URI?): 127.0.0.1:9200")
-
-
restore the ability to connect with just ip:port
If uri fails to parse, just use the connect string as given.
-
For testing the thrift transport
-
-
use MultiJson instead of Yajl directly
This increases platform compatibility and future proofs against better json libraries than yajl (like oj). Rails is using MultiJson so it isn't likely to go away. Both versions of the MultiJson api are supported. fixes #35
-
-
The old rack escape doesn't work with utf8 multibyte strings in ruby 1.9. Relying on Faraday's popularity to take care of that. If Faraday doesn't work out, Rack's new escape is supposed to be pretty good. This commit also removes support for ruby 1.8.6. Faraday only supports 1.8.7+, so the rubberband dependency is bumped too.
-
Patron relies on a C extension so it is not compatible with JRuby. Faraday supports multiple backends so it will work on more platforms. fixes #36
-
-
-
-
-
-
allow transport to be an object
This allows users to pass a transport already created in case Rubberband doesn't know how to create it properly.
-
take a connection block for transports that use it
Faraday can take a block on connect to configure middleware and connection options. Now ElasticSearch.new takes a block and passes it down to the transport also. The block will only be executed when the transport connects. This might not be right away and it might happen multiple times.
-
-
-
use DummyTransport for alternate transport tests
Travis doesn't have thrift support in its elasticsearch.
-
-
-
-
-
Add version to faraday gemspec dependency.
Fixes error when calling previously private methods.
-
-
-
respect the first argument to get_aliases
Previously this was ignored. Fixes #39
-
This overrode :size given in the query. Elasticsearch defaults to size 10 anyway so there's no need for a default here.
-
Fixing issue with ElasticSearch::RetryingClient trying forever and no…
…t respecting the @retries variable
-
-
-
MultiJson and Faraday allow running on JRuby, yay!
-
use * instead of _all to search all indices
Elasticsearch 0.19.9 no longer supports the _all syntax, and * seems to be supported at least as far back as 0.14.2.
-
set default index and type in readme
The code doesn't work as given without this. Fixes #43
-
-
-
Headers are not used by Elasticsearch, but may be used by Faraday to …
…define basic auth, so don't override them per request.
-
Checking for index existence is a common operation, and previously it was pretty cumbersome. Instead of getting the status of all indices and checking for the requested key, this implementation checks for the status of the requested index only and rescues the missing exception. This saves on bandwidth if you have a lot of indices.
-
Response status is the most useful indicator of what went wrong.
-
-
allow delete_by_query without specifying index or type
Elasticsearch allows delete_by_query across multiple indices and types, so no need to require scope.
