Re: Savannah vs. Gitlab

[H. Nikolaus Schaller]

Am 08.12.2015 um 23:06 schrieb Ivan Vučica <ivan@vucica.net>:

On Tue, Dec 8, 2015 at 5:25 PM, Svetlana A. Tkachenko <svetlana@members.fsf.org> wrote:

> In my view, due to the decentralized nature of Git, using GitHub does
> not restrict anyone's freedom.

This is wrong. GitHub users run proprietary frontend scripts on their
computers. This is not ethical.

While I don't think there is a need for GitHub to be the /primary/ storage system for GS source code, I am curious about what you are attempting to object to here?

I was also wondering about that.

Who is "they" in "their computers"?

- If GitHub, you're talking to their service over an API -- you are in control of the software running on your own computer.

- If users, what is the proprietary frontend script that I am running when executing 'git pull'?

Were you perhaps talking about the _javascript_ that is downloaded and executed by your user agent? You are not required to visit the website to obtain the free software. By mailing in patches, or by hosting your own 'fork' to publish your patches, you are not required to visit the website.

And even then, _javascript_ sent over http is open source by principle. You can always inspect the code and check what it does.

Yes, you can't change it and return it to GitHub...

But you could use our own browser engine, detect any pattern in the GitHub scripts and replace them with whatever you think is better.

Are you objecting to inability to interact with issue tracker? This might be the most valid objection, but there is an API

  https://developer.github.com/v3/issues/

and one could write a free frontend. Or one could host a separate issue tracker which, to make GitHub users happier, could offer GitHub sign-in.

> Not pull requests, no issue tracking, no code review.

The main advantage of GitHub is search-ability of the repository by
potential new contributors and a mirror already does this task. Even
though it looks pretty in my view the "code review" is an unimportant
feature and it does not have to be on a web page. This is the sort of
thing I do by e-mail for other projects.

I feel strongly that good code review system for a large, mature project is anything but 'unimportant'. Email is, also, inadequate for this purpose. (And this is not the appropriate setting to expand my thoughts on this.)

See the lengthy and serious discussion Wikimedia had about which code review system to use:

  https://www.mediawiki.org/wiki/Git/Gerrit_evaluation

I'd note that GitHub's code review tools are... wanting, and at the Dublin meeting we have generally agreed that use of any such code review tool would be optional.

 

Continuing to mention GitHub in this thread is a waste of time.

Svetlana, do you believe that statement projects an appropriate attitude?

 

If
needed, there has to be a separate conversation about writing a sync
script of "something" with GitHub after the "something" is decided

I happen to agree that bi-di sync should happen _from_ 'somewhere' _to_ GitHub.

'somewhere' could be an Internet-connected RasPi (or whatever) with GNUstep on it :)

The only thing that matters is a well known URL or IP address.

(be
it leaving things as is or moving to savannah+git, if needed).

You are very restrictive in offering options: Subversion on Gna! which happens to be uni-directionally synced to github; or Git on Savannah.

Do you truly consider those the only options?

_______________________________________________
Discuss-gnustep mailing list
Discuss-gnustep@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnustep