Re: malloc: performance improvements and bugfixes

[Jörn Engel <joern at purestorage dot com>]

On Tue, Jan 26, 2016 at 01:22:08AM -0500, Mike Frysinger wrote:
> On 25 Jan 2016 19:21, Jörn Engel wrote:
> > That said, I find language like "version 2 or later" trollbait at best.
> > The paranoid in me and many other developers starts wondering under what
> > circumstances the FSF might turn evil, by any definition of evil, and
> > create a license to further their own schemes.
> 
> this doesn't retroactively change existing releases.  so any code you
> contribute today, at worst, will be under that license (LGPL-2.1+).

That is not the point.  The point is that "or later" is signing a blank
check to the FSF to create a license as they see fit and I might
dislike.

> > Copyright assignment is far far worse.  I am signing away ownership of
> > the code.  But of which code?  Everything I ever write in the future?
> > To answer this question I have to read a lot of legalese, any developers
> > favorite.  Then I have to pay a lawyer to explain the finer points to
> > me, because I may have missed them.  Next I have to pay a second lawyer
> > to judge whether the first lawyer even knew what he was talking about,
> > which sadly isn't always the case.
> 
> FSF CLA is per-project, and iirc, only like one or two pages.  i don't
> recall it being that dense.

It doesn't show up particularly prominently in Google results either.
The closest I could find in the top ten was
http://comments.gmane.org/gmane.emacs.devel/156850

If I search for "glibc copyright assignment" I don't find anything
useful in the top ten either.  Everything I do find talks about the
copyright assignment, but I don't find the form itself.  Feels a lot
like a Douglas Adams novel so far.

Imo we should try to collect the best code in the world into a couple of
high-quality repositories.  Glibc seems like the right repository for
things like a memory allocator and I really don't mind contributing.
You can have my code under the exact terms and conditions that I
received your code under, that is perfectly fair.

But if people are asked not to even look at my code, I cannot help but
wonder how this is supposed to help the free software movement or
anything else.  If I cannot find the form I supposedly have to sign, how
can I judge whether signing them is merely annoying or will harm me in
the future.

For example, could I relicense my code under a different license after
assigning the copyright to some other entity?  That matters a lot to me
and I would expect it to matter to others as well.  Why don't I find an
FAQ page where that question is answered?

The conclusion I am coming to is that people are actively discouraged
from contributing.  Whether that is by design, accident or incompetence,
I cannot tell.  But it makes me unhappy.

> i'm not trying to push you to sign a CLA ... it's certainly your choice
> and CLA's do suck.  unfortunately, this is currently what the FSF forces
> onto GNU projects.  i'm not really sure why anymore -- the busybox/linux
> cases show that GPL enforcement is possible w/only a single copyright
> holder and no project CLA.

Sadly they also show that you mostly get bad code that wasn't worth the
legal effort you invested.  Might be nice to find the GPL in the
paperwork that accompanies your new TV, but otherwise the litigation
seemed mostly useless to me.

Jörn

--
"Error protection by error detection and correction."
-- from a university class