--- Begin Message ---
Subject: |
Re: [Gnu-arch-users] Re: towards standards specifications |
Date: |
Sat, 30 Aug 2003 19:14:00 +1000 |
On Thu, 2003-08-28 at 19:24, Miles Bader wrote:
> It's the `no more than you have to' point that seems the stickler,
> though. Tom's remark that that nothing should be getting commited
> except completely clean (heh, `pure') changesets seems like a good
> reason not to try this optimization...
Lets say in mainline you have a function "foo".
You rename to "clearer_name".
In a branch, "child", you use foo in more places than in mainline.
The merge of the change from "foo" to "clearer_name" from mainline to
child is only accurately represented by a changeset that includes
renaming the additional uses, OR, by an unaltered changeset + an
adjustment changeset.
pure-merge the script is a proof-of-concept. It immediately highlit a
problem in the design that we'd done on IRC, and led to a revised design
that is more flexible and reliable.. but it needs an enhancement to
whats-missing to allow it to work - and I've not had the time to hack up
the requisite logic. (Which is quite simple).
Rob
--
GPG key available at: <http://members.aardvark.net.au/lifeless/keys.txt>.
signature.asc
Description: This is a digitally signed message part
--- End Message ---