emacs-devel
[Top][All Lists]
Advanced

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

Re: Manual policy (Re: Emacs-devel Digest, Vol 44, Issue 65)


From: Stephen J. Turnbull
Subject: Re: Manual policy (Re: Emacs-devel Digest, Vol 44, Issue 65)
Date: Sat, 13 Oct 2007 03:09:06 +0900

Eric S. Raymond writes:

 > Stability is good.  Signs of life and responsiveness from the dev team
 > are, however, just as important.  I want to have reasonable confidence
 > that someone is alive to deal with bug reports.

It's not bug reports that are the issue in a recommendation like this
one; it's usability.  Arch is stable, and works well as designed.  The
issue I think is important is that Arch pioneered the "porcelain
vs. plumbing" distinction that git has so felicitously described; Arch
proper is plumbing, unlike what you (incorrectly) class as "more
modern" derivatives like bazaar and bzr.  They're not "more modern",
they're simply integrated and to some extent simplified (Tom would say
"dumbed-down") porcelains.  (Plumbing may be optimized, as well, which
is a huge issue with darcs and something of an issue with Arch.)

Arch has proved itself quite amenable to Emacs-based porcelains like
xetla, while still providing the raw power of Arch itself.  (Although
I haven't experienced Arch as more powerful than git; I'm curious what
use cases Manoj has in mind that Arch handles well while git does not.)

Nonetheless, I would not recommend Arch as a distributed VC to a
project looking for a VC unless the developers are experienced with
and like Arch (in which case a recommendation in vc.texi is
gratuitous, at least at *the* time of writing ;-).  As you probably
perceive, I consider the porcelain/plumbing distinction as elegant and
useful, but all active VC product communities that I know well enough
to consider for my own projects (bzr, hg, git, darcs) are very firmly
in the "porcelain-attached" camp.  I suppose that there's a reason
for that; my guess is that it gives a much smoother startup, and a
confused early history can cause a lot of pain later.

I don't have time to review the texi in detail, but I would suggest
recommending Subversion for traditional centralized organization and
as a robust vehicle for timely delivery of sanctioned versioned source
code, the foursome above (possibly adding Monotone about which I know
nothing) for general distributed use, and mentioning the existence of
GNU Arch (which after all is a GNU project and does have some special
features, not to mention substantial historic interest).  Then point
to one or more of the excellent comparison pages out on the web, or to
Wikipedia which though of uneven quality is likely to be updated
reasonably often.

HTH





reply via email to

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