Re: Current master non-deterministic

[Han-Wen Nienhuys]

On Sun, Sep 16, 2012 at 5:06 AM, David Kastrup <address@hidden> wrote:

>>> in order to a "stable" O(n lg n) uniq for which the structure of the
>>> final list does not depend on the memory order of the original
>>> elements.
>>
>> This certainly sounds like a better solution.
>
> However, rethinking this, it seems like a "solution" for not seeing the
> symptoms of a problem: we stabilize results artificially.

rather than artificial stabilizing, you could explicitly destabilize
them, so problems become more apparent. Insert from a random starting
point into the hash table, and then serialize the hash table.

-- 
Han-Wen Nienhuys - address@hidden - http://www.xs4all.nl/~hanwen