monotone-devel
[Top][All Lists]
Advanced

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

Re: [Monotone-devel] next release


From: Nathaniel Smith
Subject: Re: [Monotone-devel] next release
Date: Tue, 28 Dec 2004 15:55:16 -0800
User-agent: Mutt/1.5.6+20040907i

On Mon, Dec 27, 2004 at 02:15:37PM -0500, graydon hoare wrote:
> hi,
> 
> I'm going to push a 0.16 release out this week (I have some vacation 
> time). there's been so much work done by others during this release 
> cycle -- thank you very much! -- I actually don't know whether we're 
> "stable" at the moment or if we're waiting for some major feature to land.

I think the bug where concatenating [a] -> [b] with [b] -> [a] gives
[a] -> [a] is a release blocker.  (This is test 73.)

>   - rechangesetify my database and the one on venge.net, and tell
>     everyone that my server will panic when communicating with them
>     unless they do the same (when they send nonsense revs).

The problem with this is that it will modify any revision that
contains a rename, not just invalid revisions.  This means that if
someone has a history with renames but no invalid revisions, they will
be able to successfully netsync.  However, having done so, they'll end
up with two big parallel copies of history, diverging at the first
rename.

I see two solutions:
  - Easy, robust: just bump netsync's version number, that's basically
    the effect we want anyway
  - Harder, less robust, but perhaps shakes things up less: make sure
    the rechangesetify stuff only touches revisions that are actually
    broken
My tendency is towards the former, even if it is considered somewhat
passe to have a high protocol version number. :-)

-- Nathaniel

-- 
Eternity is very long, especially towards the end.
  -- Woody Allen




reply via email to

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