monotone-devel
[Top][All Lists]
Advanced

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

[Monotone-devel] Re: linus talk on git


From: Bruce Stephens
Subject: [Monotone-devel] Re: linus talk on git
Date: Fri, 18 May 2007 12:52:06 +0100
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.0.99 (gnu/linux)

"J Decker" <address@hidden> writes:

> Unforutnatly, although monotone has become much much better for
> performance issues... it was knocked out of the world of 'valid'
> distributed version control systems because of speed.

Yes.  IIRC, in Linus's case it was bad luck: monotone was *really*
(and unusually) slow at that point.  But he has a point: git feels
*really* fast.  Even on Windows, the MinGW port seems reasonable
(slower than on GNU/Linux, but faster than anything else available).

There are definitely things to like about git (other than speed).  Not
the weird rants about the unimportance of merge algorithms and things,
but the transient (and local) nature of branch names feels nice.

The main feature of monotone that bothers me is the issue of branch
names (including what I'd like to be temporary ones) being permanent
(and unchangeable).  Mercurial's approach of a branch being a
workspace doesn't suit me, but git's branch model seems a good one.
(If I were sure Mercurial's newish local branches were the same then
they'd also qualify, but I think they're not the same.)

Policy branches will fix this somewhat.  But still, git's branches
feel more disposable.  I guess it's a different philosophy: git allows
stuff to be disposed of, if it's local, and monotone mostly doesn't.

> There may be some issues with 'guarantee that what I put in I get
> back out' also, in respect to line endings.... I know there was some
> discussion about monotone being 'helpful' with line ending
> translation, which it really shouldn't do...
>
> There may be reasons that no matter what platform I check a certain
> text file out on that the line endings need to be exactly what I put
> in.

There may be, but my guess is that doing basically what CVS does is
probably what people want most of the time: you get the platform's
default line-endings, unless the file's binary.  (This is one of the
few things that subversion gets right, I think.  Just copying what
subversion does seems like a reasonable strategy, IMHO.)

[...]

> Git's command set appears, on the surface, to resemble
> monotone's.... GPM is one such project that has become involved with
> 'git' but it's very unclear how to actually use git to get these
> sources... maybe it's cause I started with that specific project...

I think just

git clone http://unix.schottelius.org/git/gpm

That should work, but doesn't.  Looks like the repository format is
too old or something.  Ah,
<http://unix.schottelius.org/git/HEADER.html> gives examples using
cogito, which reinforces that---I don't think anyone recommends cogito
any more.  I've no idea when the repository format changed
incompatibly, so I don't know what version of git might work.  Or
maybe cogito has always been different in this way.

And cg-clone doesn't work: some objects are missing.  Just broken, I
guess.

[...]





reply via email to

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