duplicate dependencies in Gemfile.lock on bundle update

[embark]

Error Report

Questions

What did you do?
I ran the commands:

• with the Gemfile and environment listed below
• bundle
• gem uninstall responders
• bundle update easypost

What did you expect to happen?
I expected Bundler to create a Gemfile.lock after bundle update easypost that was identical to the Gemfile.lock right after bundle

What happened instead?
Instead, Bundler duplicated responders in the gemfile:

...
     responders (2.3.0)
       railties (>= 4.2.0, < 5.1)
+    responders (2.3.0)
+      railties (>= 4.2.0, < 5.1)
     rest-client (2.0.1)
       http-cookie (>= 1.0.2, < 2.0)
...

NOTE: this only happened when using a block with the source, not if the source was just listed at the top of the Gemfile.

Have you tried any solutions posted on similar issues in our issue tracker, stack overflow, or google?
Couldn't find any similar issues, but might have missed it.

Environment

Bundler   1.14.6
Rubygems  2.5.2
Ruby      2.3.4p301 (2017-03-30 revision 58214) [x86_64-darwin16]
GEM_HOME  ~/.rvm/gems/ruby-2.3.4
GEM_PATH  ~/.rvm/gems/ruby-2.3.4:~/.rvm/gems/ruby-2.3.4@global
RVM       1.27.0 (latest)
Git       2.11.0
Platform  x86_64-darwin-16

Bundler settings

build.puma
  Set for the current user (~/.bundle/config): "--with-opt-dir=/usr/local/opt/openssl"
build.eventmachine
  Set for the current user (~/.bundle/config): "--with-opt-dir=/usr/local/opt/openssl"

Gemfile

Gemfile

source 'https://rubygems.org/' do
  gem 'easypost'
  gem 'responders'
end

Gemfile.lock

GEM
  remote: https://rubygems.org/
  specs:
    actionpack (5.0.2)
      actionview (= 5.0.2)
      activesupport (= 5.0.2)
      rack (~> 2.0)
      rack-test (~> 0.6.3)
      rails-dom-testing (~> 2.0)
      rails-html-sanitizer (~> 1.0, >= 1.0.2)
    actionview (5.0.2)
      activesupport (= 5.0.2)
      builder (~> 3.1)
      erubis (~> 2.7.0)
      rails-dom-testing (~> 2.0)
      rails-html-sanitizer (~> 1.0, >= 1.0.3)
    activesupport (5.0.2)
      concurrent-ruby (~> 1.0, >= 1.0.2)
      i18n (~> 0.7)
      minitest (~> 5.1)
      tzinfo (~> 1.1)
    builder (3.2.3)
    concurrent-ruby (1.0.5)
    domain_name (0.5.20170404)
      unf (>= 0.0.5, < 1.0.0)
    easypost (2.7.0)
      multi_json (>= 1.3.0)
      rest-client (>= 1.4)
    erubis (2.7.0)
    http-cookie (1.0.3)
      domain_name (~> 0.5)
    i18n (0.8.1)
    loofah (2.0.3)
      nokogiri (>= 1.5.9)
    method_source (0.8.2)
    mime-types (3.1)
      mime-types-data (~> 3.2015)
    mime-types-data (3.2016.0521)
    mini_portile2 (2.1.0)
    minitest (5.10.1)
    multi_json (1.12.1)
    netrc (0.11.0)
    nokogiri (1.7.1)
      mini_portile2 (~> 2.1.0)
    rack (2.0.1)
    rack-test (0.6.3)
      rack (>= 1.0)
    rails-dom-testing (2.0.2)
      activesupport (>= 4.2.0, < 6.0)
      nokogiri (~> 1.6)
    rails-html-sanitizer (1.0.3)
      loofah (~> 2.0)
    railties (5.0.2)
      actionpack (= 5.0.2)
      activesupport (= 5.0.2)
      method_source
      rake (>= 0.8.7)
      thor (>= 0.18.1, < 2.0)
    rake (12.0.0)
    responders (2.3.0)
      railties (>= 4.2.0, < 5.1)
    rest-client (2.0.1)
      http-cookie (>= 1.0.2, < 2.0)
      mime-types (>= 1.16, < 4.0)
      netrc (~> 0.8)
    thor (0.19.4)
    thread_safe (0.3.6)
    tzinfo (1.2.3)
      thread_safe (~> 0.1)
    unf (0.1.4)
      unf_ext
    unf_ext (0.0.7.4)

PLATFORMS
  ruby

DEPENDENCIES
  easypost!
  responders!

BUNDLED WITH
   1.14.6

[xaviershay]

I was able to repro. FWIW, running bundle again afterwards reverts the extra responders line.

[segiddins]

Thanks!
The fix is as follows (but will require a test):

diff --git a/lib/bundler/spec_set.rb b/lib/bundler/spec_set.rb
index 5fd6bd606..c9b53a00c 100644
--- a/lib/bundler/spec_set.rb
+++ b/lib/bundler/spec_set.rb
@@ -110,7 +110,7 @@ module Bundler
     def merge(set)
       arr = sorted.dup
       set.each do |s|
-        next if arr.any? {|s2| s2.name == s.name && s2.version == s.version && s2.platform == s.platform }
+        next if arr.any? {|s2| s.full_name == s2.full_name }
         arr << s
       end
       SpecSet.new(arr)

(the root cause being I'm guessing a spec has a platform of nil and another of Gem::Platform::RUBY)

[segiddins]

#5600 should fix this, thanks for the thorough bug report!

[embark]

Wow thank you for the super fast fix!