Re: [PATCH 10/11] push: flip !triangular for centralized

[Felipe Contreras <felipe.contreras@gmail.com>]

Elijah Newren wrote:
> On Fri, May 28, 2021 at 1:10 PM Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
> >
> > Git is a *distributed* version control system, centralized workflows are
> > the uncommon case, where we do need to do extra checks.
> 
> The commit message seemed slightly funny to me, though I'm not sure
> why.

I mean, in my mind I have present two facts:

  1. Linus Torvalds explained that the whole point of Git was to have a
     decentralized VCS [1]. It's not just an important aspect; it's
     crucial.

    If we never get past [the part where I explain distribution], I'd be
    fine with that...

    If you are not distributed, you are not worth using. It's that
    simple.

    I think this is a huge issue and that *alone* should mean that every
    single open source [project] should never use anything but a
    distributed model...

  2. Git doesn't actually have great support for "triangular workflows".

So I do find it funny that the boolean is called "triangular", when
that's supposed to be the whole point from the creator.

Maybe some of that spills out into the rhetoric of the commit message.

> ...I think the code is slightly easier to read and reason about since
> it removes the double negative.  In particular, when someone reading
> the code sees !triangular, and doesn't know or remember the meaning,
> they have to translate that to !(remote != remote_get(NULL)).
> centralized and !centralized do not have that same problem.  So, I
> like the newer version.

To my mind it's not a slight thing at all. Every time I read if
(triangular) I have to stop and translate... Oh, they mean using git in
the way it was the whole point of starting the project... So they
actually mean not using it in that *other* way.

Cheers.

[1] https://www.youtube.com/watch?v=4XpnKHJAok8

-- 
Felipe Contreras