Oh, yes, when I first hit SPRNG (actually several weeks ago, when for
whatever reason I was thinking about this stuff again), I did have the worry
that there might be GPL problems, and I was not sure whether that would then
make it something optional like NetCDF, or what. My general preference is
to get a deeper understanding of the underlying principles, do something
that is effectively published as an algorithm and in the public domain but
not code, and thereby provide the users with a solid consistent
implementation of something that everyone can easily get/use. Like I say, I
want to really think about this stuff a while; this issue has been brewing
for a couple of years, I have mentioned it as a major bottleneck, and no one
has leaned on me about it. I did not jump all over it because my impression
was that folks would be using ntt 3 for short thermal equilibration, where
high scaling is less of an issue (plus the fact that the crew out here
really does none of this stuff - I am probably the most interested party).
Regards - Bob
----- Original Message -----
From: "Ross Walker" <ross.rosswalker.co.uk>
To: <amber-developers.scripps.edu>
Sent: Wednesday, May 07, 2008 5:30 PM
Subject: RE: amber-developers: sprng parallel rng's
> So could we not just redistribute SPRNG in ambertools and then have a
> switch
> in the main AMBER configure file that links to it in place of our current
> random number generator should the user want to do that? I.e. you need
> amber
> tools and amber to build a copy of amber that links to a GPL library?
>
> Seems this would be acceptable???
>
>> -----Original Message-----
>> From: owner-amber-developers.scripps.edu [mailto:owner-amber-
>> developers.scripps.edu] On Behalf Of David A. Case
>> Sent: Wednesday, May 07, 2008 2:12 PM
>> To: amber-developers.scripps.edu
>> Subject: amber-developers: sprng parallel rng's
>>
>> On Wed, May 07, 2008, Ross Walker wrote:
>>
>> > We should probably consider using something like the SPRNG library:
>> >
>> > http://sprng.cs.fsu.edu/
>> >
>> > It is also open source so could be included in the amber tree.
>>
>> !!!NOT!!! The sprng libraries are distrubuted under the GPL, and
>> *cannot*
>> be included in Amber.
>>
>> People: please be careful about licenses...the term "open source" may be
>> misleading, but take the time to learn about this stuff. Don't borrow
>> code
>> from somewhere else unless you are sure about your right to do so. In
>> particular, GPL software cannot be linked into Amber codes.
>>
>> [We have even had instances where Numerical Recipes code was appearing in
>> places: the license for that is so restrictive that you can't probably
>> even
>> use it for personal use, much less redistribute it.]
>>
>> [One of the benefits of taking AmberTools to GPL is that we *could* build
>> on
>> things like sprng in that code base. But Amber itself has a very
>> different
>> license, and a different set of restrictions.]
>>
>> ....dac
>
>
Received on Sun May 11 2008 - 06:07:16 PDT
