[
https://issues.apache.org/jira/browse/THRIFT-231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12657625#action_12657625
]
Kevin Clark commented on THRIFT-231:
------------------------------------
>From the rdoc on Object#eql? and friends:
Equality---At the +Object+ level, +==+ returns +true+ only if _obj_
and _other_ are the same object. Typically, this method is
overridden in descendent classes to provide class-specific meaning.
[snip]
The +eql?+ method returns +true+ if _obj_ and _anObject_ have the
same value. Used by +Hash+ to test members for equality. For
objects of class +Object+, +eql?+ is synonymous with +==+.
Subclasses normally continue this tradition, but there are
exceptions. +Numeric+ types, for example, perform type conversion
across +==+, but not across +eql?
Thrift structs can't be equal unless they're of the same class. Seems odd to
take the check out.
instance_variable_get will *almost* definitely return nil for fields they don't
have in common... but how about something like:
class Foo
def initialize(msg)
@message = msg
end
end
If Foo and a thrift exception like this exist:
exception Xception {
1: string message,
}
Equality will work. Which, though unlikely, isn't unheard of.
> Ruby generated structs lack a hash method, and thus cannot be used as map
> keys or set members
> ---------------------------------------------------------------------------------------------
>
> Key: THRIFT-231
> URL: https://issues.apache.org/jira/browse/THRIFT-231
> Project: Thrift
> Issue Type: Bug
> Components: Compiler (Ruby), Library (Ruby)
> Reporter: Bryan Duxbury
> Assignee: Bryan Duxbury
> Attachments: thrift-231.patch
>
>
> Since generated structs don't define a hash method in Ruby, it defaults to a
> unique number every time. This means that even if two objects have equal
> values, they will not have equal hashcodes. As a result, if you try to create
> a set of these structs and subsequently attempt to test membership on an
> equal struct, it'll always return false. The same problem prevents structs
> from being used as map keys.
> At the very least, we should define a hash method that always returns 0 and
> leave it up to the chaining to resolve the problem. A slightly more
> appropriate move would be to have the hashcode of a struct be the composition
> of its member values.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
