gnulib-tool-py
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnulib-tool-py] Welcome


From: Stefano Lattarini
Subject: Re: [gnulib-tool-py] Welcome
Date: Fri, 27 Apr 2012 12:19:35 +0200

On 04/27/2012 11:47 AM, Bruno Haible wrote:
> Hi Stefano,
> 
>> Wouldn't it be better if the development took place in a dedicated branch
>> (or set of branches) in the gnulib official repository on Savannah?
> 
> It was my decision to have the development take place in a separate git
> repository, for several reasons.
>
OK.  Since I do have a GitHub account, I've cloned Dmitriy's repository
there, so that it will be much easier for me to observe the development
if I want to, without requiring extra actions from Dmitriy.

> One of the reasons is that Dmitriy shouldn't
> be constrained with file naming and directory naming conventions. The way
> the rewrite is integrated in the end (which file names, which directory
> names) is something that is better decided at the end.
>
Makes sense.  After all, in a rewrite from-scratch, one doesn't need to
keep the repository clean and "complaint" all the time.

> And, as you know, file renames get in the way of git history;
>
No, they don't; git is extremely good at detecting renames.

> at this point a move of the code to gnulib's main repository and
> 'master' branch will not be much of a problem.
>
You mean you are going to lose the history of Dmitry's repository once you
merge his work in the Gnulib official tree?  That seems awful (especially
because his commit messages seems clean and useful, even at this early
stage) ...

Update: Oh, I guess you'll be using a "subtree merge" not to lose the new
history: <http://nuclearsquid.com/writings/subtree-merging-and-you/>.
I wasn't aware of such a possibility until now.  Git never ends to amaze
me :-)

>> In addition, I believe most gnulib contributors don't even have a GitHub
>> account (even Bruno doesn't, in fact, unless I'm mistaken).
> 
> Dmitriy should be the only committer to his code. For contributing to the
> wiki, yes, I will need an account at Github.
>
My point was that the most useful features of GitHub (simple bug tracking
through issue reporting, pull requests, repository observation) won't be
available to you unless you have a GitHub account.  But if that's fine for
you, I certainly won't complain -- especially because I can use those
features :-)

>>> so you can see the development process and all the changes.
>>>
>> I think the policy around here it to send each change to the mailing list for
>> review before applying it; even if it is a quick fix or an uncontroversial
>> change that can be applied right away, it should still be sent to the list
>> as an "FYI".
> 
> What you describe are the policies for gnulib's repository. Here Dmitriy's
> project will be full-speed. He does not need to ask anybody. If I disagree
> with some of his choices, I will tell him; that's part of my duty as "mentor".
>
OK (even if I think that more eyes are always better).  And anyway, I trust
Dmitriy that we will seek community counsel before taking important and
far-reaching decisions (i.e., whether and how to support Python 3, or which
testing library to use), so not big deal.

Thanks for clarifying the matter,
  Stefano



reply via email to

[Prev in Thread] Current Thread [Next in Thread]