Boost logo

Boost :

Subject: Re: [boost] [xint] Third release is ready, requesting preliminary review
From: Chad Nelson (chad.thecomfychair_at_[hidden])
Date: 2010-05-02 13:22:19


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 05/02/2010 12:37 PM, Scott McMurray wrote:

>> I've been assuming that the copy-on-write was responsible for the
>> major part of the speed difference, but after thinking about it,
>> that doesn't make a lot of sense. The serialized access to the
>> random number generator could be responsible for a good part of it.
>> The tests I based that assumption on used it heavily. I'll run some
>> further tests when I have a chance today to see how they turn out.
>
> Why do you have a global random number generator at all?

Because I hadn't thought about an alternative until you said this. :-)

> Boost.Random doesn't provide one, so I don't see why you should either.

It *would* simplify things greatly for me. I'm just concerned that it
might make things more difficult for someone who just wants to use the
library for a quick one-off test. My goal is, as much as possible, to
stay out of the way of the person using the library -- for that, having
a default "good enough" random generator is pretty much a necessity.

(The use-case I'm basing that on: When I'm trying to solve a problem,
and go hunting for a tool that I need to do it, it's frustrating to have
to spend a lot of time learning how to set up the tool -- I just want
something that I can quickly use to solve the problem at hand. Later,
when I'm not so focused on a specific problem, I'll gladly dig into the
documentation and figure out the non-default features that I might find
a future use for.)

Maybe if I had a random number generator as a parameter, defaulting to
an internal global one that is deliberately *not* thread-safe, and
simply documented that if you use it from multiple threads, you're
responsible for either serializing the access or providing your own
generators...?
- --
Chad Nelson
Oak Circle Software, Inc.
*
*
*
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkvdtMsACgkQp9x9jeZ9/wSR0gCeNZ1DAFzFnmz9SPT7dmvgw3/Y
/cIAoMmZQFMWfxLTQBQSW9Okerxsdz7f
=qEQ6
-----END PGP SIGNATURE-----


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk