Richard Fontana wrote:
2. I continue to think the strongest argument against use of GitHub
for
copyleft-next is the one you originally made long ago:
https://github.com/richardfontana/copyleft-next/issues/6#issue-5447013
Two things have changed for me since I wrote the text in that link above.
First, the Gitoriocalypse. Gitorious was AGPLv3'd, but it *clearly did not
respect net services freedom* because there was no way to export and move to
your own instance of the same software. In other words, having the SaaSS
available under Affero GPL is a *necessary* but clearly and undeniably not a
*sufficient* condition for net.services freedom. There was a day last year
when I was ironically wishing my Gitorious projects had been on GitHub
because at least I'd have an API to extract my data!
Second, failure by Free Software zealots to convince any but a very few
number of developers to realize the dangerous situation with GitHub.
I'm starting to wonder if a project using GitHub with a culture of regularly
discussing *why* GitHub is a horrible compromise and why it is a problem
might be more optimal strategy. (I'm thinking something akin to "using Unix
machines to write GNU" sort of mentality.)
Specifically, GitHub is so popular now that virtually any project refusing to
use it will fail to make a dent in GitHub's popularity, but will take a huge
hit in their own. (i.e., the Linux and Git projects themselves refuse to use
GitHub, but I bet if we polled GitHub users, 99% of them wouldn't even know
that, and even Linux and Git maintain read-only mirrors on GitHub). I'm thus
convinced, at a minimum, that it's an absolute necessity to interact GitHub
(in a Free Software only way, which a 'git push' certainly allows) to publish
read-only copies of projects. It's an important way to reach the GitHub
community to let them know your project exists, *and* likely the very best
way to get the message out that "not everyone uses GitHub, and there are good
reasons not to!". And, to bring this back to our situation at hand:
copyleft-next is always going to at least have a read-only mirror of its Git
repository on Github.
Meanwhile, if it's true that GitHub's issue tracker and pull requests can be
interacted with fully through a Free Software command line program (which is
a huge change since my original comments about coyleft-next being on GitHub),
it might be possible to separate out the question of proprietary clients.
Thus, there are three distinct (but interrelated) issues with GitHub:
A. The SaaSS truly server-side code, that will probably never be released.
B. The mixed SaaSS+proprietary Javascript web interface.
C. Other Franklin-Street-Statement-related issues, such as data
portability, federation, etc.
GitHub is abysmal on A, problematic-but-regularly-getting-better on C. And,
B can almost be ignored entirely because their improvements on C allows for
Free Software replacements for B. (And, under the "don't contribute to a
proprietary program's popularity" rule, we're now in the "GNU binaries
for
Microsoft Windows" territory: it's not like providing them caused someone to
use Microsoft Windows who wasn't going to do so anyway, and the same is true
with virtually any but the biggest projects in the world for GitHub.
copyleft-next's use of GitHub will presumably not cause anyone to sign up for
a GitHub account [0].)
Now, mix in the fact that GitHub has culturally succeed at using "Open
Source" messaging to gather (in its social networking aspects) just about
everyone who might even tangentially contribute to Open Source and Free
Software into a user base. (Even I have an active GitHub account, and I'm
actively boycotting Twitter, Linked-In, and Facebook.)
I think we never faced a situation quite like this one in the history of
software-development-infrastructure SaaSS. I used to think it's "SourceForge
all over again", but it's more complicated than that now. Boycotting
SourceForge made obvious sense and it clearly succeeded as a strategy, but it
was really stalemate, because our defeat of SourceForge only left open the
possibility for a GitHub to exist. I'm dubious that a boycott of GitHub will
yield the outcome of defeating GitHub and/or reforming its
anti-software-freedom and anti-net-services freedom policies.
It is more acute today because I now believe that copyleft-next
should
address the issue of network services freedom (which I did not believe
when I started the project).
I almost did a little dance here in my apartment at reading that. :) I'll
get you some old patches back RSN. :)
But, as I said above, I think an Affero-like clause is definitely a necessary
condition for future net-services freedom, but it's not sufficient. I doubt
we could construct something we'd consider an Open Source and Free Software
license that could handle the other Franklin Street Statement issues (but it
is possible I'm just not thinking creatively enough).
3. Until recently I have had practical difficulties in setting up
self-hosted git-based web service things I've investigated (Kallithea,
Gitlab, Pagure, Gogs ...).
This is a relevant problem for sure. The Kallithea instances I'm in charge
of, which are only a few, require substantial attention from me to keep
running.
[0] I almost reworked my entire argument, though, when I considered that
lawyers who don't like Git might actually contribute to copyleft-next if
it eventually becomes anything more than a toy project, in which case it
*would* draw new users to GitHub. That's something to think hard about.
--
-- bkuhn
========================================================================
Become a Conservancy Supporter today:
https://sfconservancy.org/supporter