Re: On the subject of Git, Bazaar, and the future of Emacs development

[Eli Zaretskii <eliz@gnu.org>]

> From: chad <yandros@MIT.EDU>
> Date: Tue, 2 Apr 2013 01:25:41 -0700
> Cc: Barry Warsaw <barry@python.org>, "Stephen J. Turnbull" <stephen@xemacs.org>
> 
> On 01 Apr 2013, at 20:24, Stephen J. Turnbull <stephen@xemacs.org> wrote:
> 
> Bzr itself is Not Simple, and until it effectively stagnated, was
> also not especially stable.  The internal models are relatively
> complex, allowing bazaar to handle many different workflows, and
> the structures (for example, the repository layout) were changing
> relatively frequently.  When Emacs adopted bazaar, bzr was on the
> cusp of a big change, with another on the horizon (looms was one
> candidate; I've forgotten the name of the other).
> 
> As things slowed down inside bzr, the barriers to entry went up,rather
> than down. Patches sat around bit-rotting rather than being included
> or rejected, which made it hard (at least conceptually) to get up
> to speed with the project. At the same time, the core thinned (Martin
> Pool, for example).

That reminds me of another project I'm familiar with: Emacs.

It is definitely "Not Simple", and the development versions many
people use every day are not terribly stable, either.  "Complex
internal models"?  we've got more than bzr could ever dream of.  I'm
hacking the display engine since 2008, and it still surprises me from
time to time.  "Frequent changes in structures"? you betcha, just look
at the logs from the last month.  "Thin core"? the guy who implemented
the display engine is no longer with us, since 10 years ago, and the
couple of others who knew a lot about redisplay are not active for at
least 5 years.

If I were reasoning like you do, I'd never have written the
bidirectional display code.  Why did I?  Because (1) it was a feature
I lacked in Emacs and knew I would use when it's available, and (2) it
pissed me off to have to use those "other tools" whenever I needed
this feature.  IOW, I was motivated, and also experienced enough in
the programming paradigms required to do the job, however hard (it
eventually took me 2 full years).

Doesn't this remind you something?