Bug filed: Array#hash should properly handle recursive arrays


#1

Hi,

I have just filed this bug on CodePlex:

It is actually pretty easy to fix with a bunch of lines of code in
RubyArray.GetHashCode (just tested, too bad IronRuby.dll is not open
to contributions) and, as a result, this would resolve the following 4
critical failures for the array core specs:

Array#hash properly handles recursive arrays
Array#| properly handles recursive arrays
Array#uniq properly handles recursive arrays
Array#uniq! properly handles recursive arrays

Regards


Daniele A.
http://www.clorophilla.net/blog/
http://twitter.com/JoL1hAHN


#2

I can fix RubyArray.GetHashCode to handle the simple case of an array
containing itself.

However, the simple fix wont deal with the case where the array is
indirectly recursive as in this snippet.
a = []
b = [a]
a<<b


#3

Curt pointed out a way to fix the recursive case using
RubyUtils.RecursionTracker…


#4

Great, that’s exactly what I used :slight_smile:

Thanks

On Fri, Apr 17, 2009 at 00:22, Shri B. removed_email_address@domain.invalid
wrote:

However, the simple fix wont deal with the case where the array is indirectly recursive as in this snippet.
Hi,
Array#| properly handles recursive arrays
Ironruby-core mailing list
removed_email_address@domain.invalid
http://rubyforge.org/mailman/listinfo/ironruby-core


Daniele A.
http://www.clorophilla.net/blog/
http://twitter.com/JoL1hAHN


#5

Does Ruby call “hash” dynamically on each element of the array?
If not then there is no need for RecursionTracker. A simple
non-recursive DFS with a dictionary that remembers visited objects
should do.

Tomas


#6

See RubyFileOps.Join for an example.

Tomas


#7

The simple test I did suggested that MRI dynamically calls hash on each
element of the array.