Subject: Re: [boost] [1.47.0] Any remaining showstoppers?
From: Neal Becker (ndbecker2_at_[hidden])
Date: 2011-07-07 08:04:54
Beman Dawes wrote:
> On Tue, Jul 5, 2011 at 11:04 AM, Steven Watanabe <watanabesj_at_[hidden]> wrote:
>>>> I have 2 Random fixes:
>>>> Both of these remove interface breaking changes.
>>> Were these supposed to fix TR1 run_random and run_std_random tests?
>>> They still seem to be failing.
>> Those tests didn't even compile before.
>> The failures now are a result of the changes
>> in the seeding algorithm. The old seeding
>> algorithms were bizarre and inconsistent,
>> so I really don't want to go back.
> OK, understood. Please go ahead and merge those changesets.
> Is anyone working on changing the TR1 tests to match the seeding?
> Should you open a ticket?
Change to seeding? Does this mean that random sequences generated with the same
seeds as used pre-1.47 will be different? Does this mean that all my tests that
used random number generators will now fail to match previous results? If so,
please say something in release notes.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk