From: "akr (Akira Tanaka)" <akr@...>
Date: 2013-11-08T01:19:24+09:00
Subject: [ruby-core:58219] [ruby-trunk - Bug #9089] rb_fix2uint no longer	raises a RangeError when given negative values


Issue #9089 has been updated by akr (Akira Tanaka).


One more thing:
FIX2UINT(-1) doesn't raise an exception on 32bit environment on Ruby 2.0.
----------------------------------------
Bug #9089: rb_fix2uint no longer raises a RangeError when given negative values
https://bugs.ruby-lang.org/issues/9089#change-42802

Author: NoKarma (Arthur Schreiber)
Status: Feedback
Priority: Normal
Assignee: 
Category: core
Target version: current: 2.1.0
ruby -v: ruby 2.1.0dev (2013-11-07 trunk 43560)
Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN


Up until the change that was made in ((<URL:https://github.com/ruby/ruby/commit/92f59c6d7937b14bb5eefb052099ef0a3ef3bcd0>)), (({rb_fix2uint})) would raise a (({RangeError})) if it was given a negative value like (({-1})) (e.g. when using the (({FIX2UINT})) macro).

Due to the changes made in that commit, this error is no longer raised and (({rb_fix2uint})) will return an incorrect value instead.

There is a C-API spec in rubyspec that shows that this behavior has changed between 2.0.0-p247 and 2.1.0-preview1, and I checked and made sure this is still not working correctly in the latest 2.1.0-dev version. The failing spec can be found at ((<URL:https://github.com/rubyspec/rubyspec/blob/master/optional/capi/fixnum_spec.rb#L16-L18>)), it is part of the "optional" c-api specs.

Is there any reason why there is the (({if (num < (unsigned long)INT_MIN)})) is made inside the (({check_uint})) function? Doesn't the (({sign})) parameter automatically indicate that we can't convert to an unsigned integer?


-- 
http://bugs.ruby-lang.org/