Boost logo

Boost :

From: David Abrahams (dave_at_[hidden])
Date: 2007-05-30 20:15:24

on Tue May 29 2007, Christopher Woods <> wrote:

> Questions (from the novice/newbie/outsider):
> 1) Is it a *requirement* for any new libraries that are submitted
> for review, currently under review, or reviewed/accepted but not yet
> in the Boost distribution accept the BSL?

I'm going to go out on a limb and say "yes."
An undocumented requirement, but still...

> 2) Are there any other libraries of Boost that are dependent upon uBLAS?


> If the answers are "Yes" and "No" respectively, then I don't quite
> understand how a single special case is all that harmful to Boost.
> (Maybe because I'm not a lawyer.)

Or maybe because you don't work in a company where the lawyers can't
don't like complication, or because you can't see how having one
exception causes pressure to allow more exceptions.

> Certainly having many cases (as in before the BSL push/adoption) was
> harmful if not impossible. I understand and agree with the need for
> a single license but when you are down to 1-2 "stand-alone" cases
> then the harm to boost is fairly minimized is it not?

Reduced, but IMO not acceptable.

> Wouldn't it be sufficient to simply document this special case on
> the license information page
> ( something like:
> _*Non-BSL Conforming [Legacy] Libraries*_
> uBLAS currently remains the only library included with the Boost
> distribution that has not adopted the BSL (yet). Please see the
> library for licensing details.
> *No Boost libraries are dependent upon any non-BSL library in any
> manner - nor will they ever be. All new libraries added to Boost
> are required to accept the BSL.*

We may have to do that if we can't get the uBlas authors to change the
license and we can't come to a different consensus, but I would not
like it. It's a wart, and complicates makes legal analysis of the
implications of using Boost.

> Something like that should allow companies to still accept Boost
> from a legal standpoint readily enough, no?

I can't speak for them. My goal is to remove the barriers to adoption,
not just keep them moderately low.

> They could review BSL, find it sufficient and then say to their
> developers "you can use Boost except for uBLAS"

In some cases, they don't trust the developers. uBLAS would actually
need to be removed from the code to which they have access.

> (if the developers don't need uBLAS) or they can spend the extra
> effort to review uBLAS's license (if/when the developers need it).
> If that's not enough, how about additionally moving uBLAS to a
> separate namespace? "boost_non_conformant", "boost_non_BSL", or
> something along those lines.

I don't see how that helps. Lawyers don't care about namespaces.

Dave Abrahams
Boost Consulting

Boost list run by bdawes at, gregod at, cpdaniel at, john at