YAOSL - Yet Another Open Source License

Alex Nicolaou anicolao at cgl.uwaterloo.ca
Fri Oct 15 05:10:58 UTC 1999

Bruce Perens wrote:
> From: Alex Nicolaou <anicolao at cgl.uwaterloo.ca>
> > You may not distribute versions of the software with your patch applied,
> > although you may include your patches in a subdirectory of the software
> > specifically provided for this purpose, or provide the software in an
> > archive format designed to accomodate and apply your patches during
> > installation on the end user's computer.
> No current open source license prohibits the distribution of modified binaries.
> I think you should require that they be marked clearly as modificatied
> versions, but you need to allow them to be distributed.

While I agree that this needs to be added to my license, I'm not sure
why modified binary versions really need to be distributed in general. 

In the case of my particular project, a lot of people don't have the
facilities to build the binaries and so the license clearly needs to
allow distribution of modified binaries. This is really a case of
providing for the convenience of the end user and developer who are
working on a forked version of the software.

> Regarding the requirement to send back modifications even if they are only
> being used internaly, does that mean I have to send them back with each
> test run and edit? I think you should tie this to _public_performance_ rather
> than internal development, or drop it.

Hmm, don't know how to resolve this. Maybe it should be dropped, maybe
not - see below.

> Regarding the sending back of modifications - it's a big hassle to send them
> back with _every_edit_ - consider the overhead on a distribution like Debian
> if they had to synchronize with every developer before every beta test they
> do. Also, you get mail-bombed with tons of useless deltas without any way to
> review or classify them because people rig their CVS servers to send you a
> delta at every check-in. What I negociated with ATT, IBM, Apple, etc. was that
> the developer would submit to them a URL where the source code for what they
> were distributing could always be found. That way, the developer only needed to
> send one email upstream, and the upstream developer (you) always had an
> better-organized place where you could see the developer's entire version
> rather than a mess of uncategorized patches.

You aren't required to send back modifications if you publish them; it's
3a OR 3b, not both. So for the Debian case there's no issue since the
modifications are publicly available anyway. This also allows for the
URL solution you discuss. Am I right in thinking that you thought it was
3a and 3b, or is there a problem even if it's 3a or 3b?


More information about the License-discuss mailing list