rubygems-2.0.3 gem install fails when a gem is found with the correct specs, but the wrong architecture

[djellemah]

There's an icorrect format string (%) in the code below:

def wordy
  "Found %s (%), but was for platform%s %s" %
    [@name,
     @version,
     @platforms.size == 1 ? 's' : '',
     @platforms.join(' ,')]
end

[luislavena]

@djellemah thanks for reporting this.

I can confirm this and created the following patch (against master)

diff --git a/lib/rubygems/errors.rb b/lib/rubygems/errors.rb
index fdbe39b..e296ef3 100644
--- a/lib/rubygems/errors.rb
+++ b/lib/rubygems/errors.rb
@@ -62,10 +62,10 @@ module Gem
     ##
     # A wordy description of the error.
     def wordy
-      "Found %s (%), but was for platform%s %s" %
+      "Found %s (%s), but was for platform%s %s" %
         [@name,
          @version,
-         @platforms.size == 1 ? 's' : '',
+         @platforms.size == 1 ? '' : 's',
          @platforms.join(' ,')]
     end
   end
diff --git a/test/rubygems/test_gem_spec_fetcher.rb b/test/rubygems/test_gem_spec_fetcher.rb
index b4aff09..a821057 100644
--- a/test/rubygems/test_gem_spec_fetcher.rb
+++ b/test/rubygems/test_gem_spec_fetcher.rb
@@ -132,8 +132,10 @@ class TestGemSpecFetcher < Gem::TestCase

     assert_equal 0, specs_and_sources.size
     assert_equal 1, errors.size
+    pmm = errors.first

-    assert_equal "i386-linux", errors[0].platforms.first
+    assert_equal "i386-linux", pmm.platforms.first
+    assert_equal "Found pl (1), but was for platform i386-linux", pmm.wordy
   end

   def test_spec_for_dependency_bad_fetch_spec

Also corrected the pluralization of platforms (was incorrect).

@drbrain and @evanphx can you take a look to above patch? I'm not familiar with the code or the tests in questions so dunno if this is the best place to put it.

Thank you.

[drbrain]

@luislavena that patch looks good, please commit it!

[luislavena]

Merged at e0bbe20

[luislavena]

@drbrain any plans for a 2.0.x bugfix release? we have a lot of things to be backported 😢

[drbrain]

Once the 2.1 milestone is complete (shortly) we can start work on a 2.0.x bug fix (probably next week).