> My concern was mainly related to the fact I didn't know what they were,
which you have answered adequately. Are they LGPL?
> libpstat and fskit are both LGPLv3 (my motivation for v3 over v2 is the
explicit patent license).

Generally speaking, with the GPLv3 patent clauses, the only people who worry
about the patent clause are people who actually hold patents. In
practicality, the GPLv3 is waste of time because it does not protect you.
If you or I write GPLv3 code and it is considered infringing by a third
party patent holder, the GPLv3 will not legally prevent you or your project
handed a fully enforceable "cease and desist" order or from being sued for
damages, nor will it protect your users or downstream coders from the same
lawsuit -  because you were never granted a license in first place.  Thus,
unless you are a patent holder, writing code for use, the GPLv3's patent
clauses are not horribly useful.     If you want protection from patent
litigation, look to the OIN instead.

I remember reading that Black Duck estimated the use of GPL3 at about 12%
and the LPGL at maybe 3% out of all GPL licensing total. The Linux kernel
developers have strong objections to GPL3, and so do I. The thing that
bothers me with the GPLv3 is that it is a bad license, conceived for
political and philosophical reasons that are almost completely divorced from
the reality of law, software development in the world we live in. You might
think that I am being harsh and overcritical, but I do not believe so. Yes,
the license was examined carefully by lawyers - so what? So was the US
DMCA, and we all know how that turned out in court.

A critical example, the "additional terms" clause is designed specifically
so that one author can limit where and how opensource code can be used by
downstream programmers. Stopping the use of code defeats the entire purpose
of the license, the FSF, and open source in general. Additional terms can
be added by downstream coders to limit it even further. Opensource licenses
should encourage the use and improvement of code, not actually prevent it.

I would characterize the GPLv3 as a "potentially abusive" license that
diminishes the rights of downstream coders, while offering very little
actual patent protection, unless the upstream coder is someone like Intel or
Microsoft.


>Again, I'm willing to dual-license them with something more BSD-friendly.
I was leaning towards ISC earlier since that's what OpenBSD uses, but I
don't have a strong opinion. Is there something the MIT license offers that
you particularly like?

Why do I like the MIT license? Because it easily allows downstream
programmers to relicense as virtually any license you want. The code can be
incorporated into whatever it is needed for, even if you need it for a
closed company product. The whole point of open code is to have it be
usedand HOPEFULLY improved. You might object to the lack of copyleft in the
MIT license, but I prefer to deal with reality rather than idealism. The
reality is that there are very few real world cases where people actually
ask for code that has not already been released into the wild voluntarily.
Copyleft enforcement cases have forced some companies to release code, that
is true - but in the end what real value did forcing the issue offer the
community? The end result is that the violator stopped using GPL code sure,
but then they probably fired the people who might have contributed to
opensource as well as scaring away the management from greenlighting using
opensource in the company.

That is why I respect the GPLv2 license for what it is, I tend to lean
toward the MIT. I dual license GPLv2/MIT so that I get the best of both.
The reason I prefer this is so the code can be used by the widest audience,
and incorporated into what would otherwise be a GPL incompatible license and
yet a version is still guaranteed to be available for everyone.