summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2016-07-12Merge pull request #5 from azul/celluloidazul
Use Celluloid, Reel and Celluloid I/O based http requests
2016-07-04fix start / stop scriptAzul
I have to admit I don't completely understand what is going on here. https://github.com/celluloid/celluloid/wiki/Frequently-Asked-Questions says you can either use sleep after initiating the actors or call run on them. But this might not be true for reel. At least run did not work and now new; sleep seems to work fine.
2016-07-02refactor: turn EmailAddress into a class, cleanupAzul
2016-07-02refactor: replace blocks/yields with returnsAzul
This became possible because we now use celluloid. Celluloid handles asynchronity without the need for callbacks or blocks.
2016-07-02minor: silence warnings and cleanupAzul
2016-06-30use proper ssl connection with ca_file - green!Azul
2016-06-30actually use celluloid io for http requestsAzul
2016-06-30fix host header handling and testsAzul
2016-06-30Trying to replace EM base server with reelAzul
some tests are still broken. But at least they are running now.
2016-06-30update webmock versionAzul
2016-06-24first attempt at using celluloid instead of em - brokenAzul
Main problem right now is the outdated version of webmock.
2016-06-21refactor: change send_response -> respond in em_serverAzul
This way the interface mimics the one of Celluloid Reels request. request.respond status, content
2016-06-21Merge pull request #4 from azul/refactor/separate-em-serverazul
refactor: separate em specifics from generic server
2016-06-20seperate generic RequestHandler from EmServerAzul
The request handler takes a responder (the EmServer) and an adapter as initialization arguments. It will then respond to requests by querying the sources using the adapter and calling send_response on the responder.
2016-06-20add Nickserver::Source as superclassAzul
2016-06-20refactor: separate em specifics from generic serverAzul
2016-06-15Merge pull request #3 from azul/refactor/transport-adaptersazul
Refactor em specifics into http adapter
2016-06-15we don't need FetchKeyInfo anymore including in SourceAzul
2016-06-14introduce Hkp::ClientAzul
wraps the hkp protocol
2016-06-14use adapter for FetchKeyInfoAzul
2016-06-11use the adapter not EM in hkp sourceAzul
2016-06-10turn Hkp::FetchKey into Hkp::SourceAzul
The source was really just an empty shell now that we pushed the em specific stuff further down.
2016-06-09refactor: remove EM specific stuff from Hkp::FetchKey interfaceAzul
2016-06-09add Hkp::Source with same interface as CouchDB::SourceAzul
2016-06-09refactor: introduce Hkp::ResponseAzul
2016-06-09refactor: inline method get_key_from_uidAzul
well kind of since we were yielding this is more inlining the block and renaming the method
2016-06-09move hkp test to integration testsAzul
That's what it actually is
2016-06-09refactor: remove middleman Couch::FetchKeyAzul
2016-06-09Merge pull request #2 from azul/rubocopazul
Rubocop
2016-06-08refactor: use new couch_db classes from fetch_keyAzul
This moves all the em_http related stuff into a single adapter. We're also not using callback and errback inside fetch_key or server for couch requests anymore. Changing the interface for hkp to do the same will follow.
2016-05-25add minitest/hell to the mixAzul
This will randomize test order and parallelize them iirc
2016-05-25silence all warnings of webmock and evma_httpserverAzul
2016-05-25fix our own warningsAzul
2016-05-25silence some warnings from evma_httpserverAzul
2016-05-25move nickserver_test to integration dirAzul
it really is more of an integration test
2016-05-25simplify KeyInfo. remove unneeded accessorsAzul
2016-05-25run all tests by defaultAzul
2016-05-25copy over all files from rewritten attemptAzul
I started a nickserver from scratch to implement the things that are independent of our choice of stack (eventmachine or other). This commit copies them over and tests both things in parallel.
2016-05-25rubocop: fix unused block paramsAzul
2016-05-25rubocop: disable EndAlignment cop permanentlyAzul
we use end; end to end nested modules.
2016-05-25rubocop: fix deprecated class methodsAzul
2016-05-25rubocop: initialize and use ruby 1.9 hash syntaxAzul
2016-05-25Version 0.3.00.3.0Azul
2016-04-10travis: match production environmentversion/0.8Azul
2016-04-10test: try latest rubygems and bundlerAzul
2016-04-10test: downgrad rubygems to fix issue with bundlerAzul
2016-04-10minor tweaks to hkp response parsingAzul
Only parse responses that have status code 200 (OK). Simplify status code handling a bit Also profiled it to see if duplicate calculations matter. They don't (2ms for validating 12 keys)
2016-04-10add test for 200 without matching keysAzul
2016-04-10deprecation: File.exists? -> File.exist?Azul
2016-04-10bugfix: 404 if no key matches, fixes #6789Azul
I also separated the parsing of the hkp response from FetchKeyInfo. This way FetchKeyInfo has the EM specific code that has sideeffects and the logic is in a class without sideeffects and (almost) without state. The only state we keep is the KeyInfo array that contains all the information the server returns. This way we avoid parsing the response multiple times.