atom feed75 messages in org.freebsd.freebsd-hackersRe: The -stable problem: my view
FromSent OnAttachments
Greg LeheyJun 7, 1996 3:15 am 
Jordan K. HubbardJun 7, 1996 6:29 am 
Nate WilliamsJun 7, 1996 8:29 am 
Karl Denninger, MCSNetJun 7, 1996 8:41 am 
Nate WilliamsJun 7, 1996 10:02 am 
Terry LambertJun 7, 1996 11:22 am 
Angelo TurettaJun 7, 1996 11:48 am 
Marc G. FournierJun 7, 1996 12:05 pm 
Nate WilliamsJun 7, 1996 12:53 pm 
Terry LambertJun 7, 1996 12:53 pm 
Nate WilliamsJun 7, 1996 1:07 pm 
Terry LambertJun 7, 1996 2:12 pm 
Nate WilliamsJun 7, 1996 2:26 pm 
Terry LambertJun 7, 1996 2:58 pm 
Nate WilliamsJun 7, 1996 3:07 pm 
Terry LambertJun 7, 1996 3:15 pm 
Nate WilliamsJun 7, 1996 3:24 pm 
Jordan K. HubbardJun 7, 1996 4:00 pm 
Nate WilliamsJun 7, 1996 4:06 pm 
Jordan K. HubbardJun 7, 1996 4:27 pm 
Jordan K. HubbardJun 7, 1996 4:40 pm 
Terry LambertJun 7, 1996 4:42 pm 
Jordan K. HubbardJun 7, 1996 4:56 pm 
Justin T. GibbsJun 7, 1996 5:35 pm 
Jordan K. HubbardJun 7, 1996 5:59 pm 
Jordan K. HubbardJun 7, 1996 6:07 pm 
Nate WilliamsJun 7, 1996 6:13 pm 
Scott BlachowiczJun 7, 1996 7:03 pm 
Michael HancockJun 7, 1996 7:08 pm 
Nate WilliamsJun 7, 1996 7:21 pm 
Michael HancockJun 7, 1996 7:25 pm 
Michael HancockJun 7, 1996 7:30 pm 
David GreenmanJun 7, 1996 9:06 pm 
Nate WilliamsJun 7, 1996 9:07 pm 
Terry LambertJun 7, 1996 9:10 pm 
Nate WilliamsJun 7, 1996 9:18 pm 
Terry LambertJun 7, 1996 9:39 pm 
Terry LambertJun 7, 1996 9:58 pm 
Jordan K. HubbardJun 7, 1996 10:13 pm 
Jordan K. HubbardJun 7, 1996 10:24 pm 
Terry LambertJun 7, 1996 10:24 pm 
Nate WilliamsJun 7, 1996 10:26 pm 
Jordan K. HubbardJun 7, 1996 10:33 pm 
Nate WilliamsJun 7, 1996 10:34 pm 
David GreenmanJun 7, 1996 10:40 pm 
Terry LambertJun 7, 1996 11:03 pm 
John BirrellJun 7, 1996 11:18 pm 
Justin T. GibbsJun 7, 1996 11:22 pm 
Rodney W. GrimesJun 8, 1996 1:24 am 
Jordan K. HubbardJun 8, 1996 2:28 am 
NarviJun 8, 1996 2:33 am 
NarviJun 8, 1996 2:53 am 
NarviJun 8, 1996 3:05 am 
Heiko BlumeJun 8, 1996 4:29 am 
Michael HancockJun 8, 1996 8:22 am 
Garrett WollmanJun 8, 1996 11:18 am 
Justin T. GibbsJun 8, 1996 1:52 pm 
M.R.MurphyJun 8, 1996 2:10 pm 
Terry LambertJun 8, 1996 2:52 pm 
M.R.MurphyJun 8, 1996 3:40 pm 
Jordan K. HubbardJun 8, 1996 4:08 pm 
Julian H. StaceyJun 9, 1996 3:54 am 
Randy TerbushJun 9, 1996 7:21 am 
Bob WillcoxJun 9, 1996 7:54 am 
Chuck RobeyJun 9, 1996 8:34 am 
Randy TerbushJun 9, 1996 8:40 am 
Ollivier RobertJun 9, 1996 12:56 pm 
Bob WillcoxJun 9, 1996 2:32 pm 
Andrew V. StesinJun 10, 1996 1:45 am 
NarviJun 10, 1996 6:54 am 
Andrew V. StesinJun 10, 1996 7:40 am 
NarviJun 10, 1996 8:48 am 
Michael RobinsonJun 14, 1996 2:38 am 
Amancio HastyJun 14, 1996 3:00 am 
Terry LambertJun 17, 1996 11:32 am 
Subject:Re: The -stable problem: my view
From:Nate Williams (na@sri.MT.net)
Date:Jun 7, 1996 4:06:27 pm
List:org.freebsd.freebsd-hackers

doing a *ton* of work in both -stable and -current. However, it's a *LOT* of work. However, I don't think this has anything to do with CVS, but has to do with the diverging of the trees. P3 may make it easier to do as far as resources, but the actual work of 'merging' in changes to both won't be any easier. Building the patches is the hard work IMHO,

I think you're forgetting the problem with cvs where:

1. You make a change in -release. 2. You merge it into -stable. 3. You make another change in -release. 4. You go to do another merge into -stable and wind up with a whole *mess* of conflicts. `cvs update -j' is NOT a decent merge tool!

I don't use 'cvs update -j' to merge it into -stable. 'merge' doesn't work when you've already merged. What *I* do is this, which is a bit more work but does do the job 95% of the time.

% cvs log file-to-merge. (Figure out which revision(s) needs to go into the branch). % cvs diff -c -r1.A -r1.B file-to-merge > cdif % cvs update -r BRANCH_TAG % patch file-to-merge < cdif .... % emacs -nw file-to-merge file-to-merge.rej .... % cvs diff -bu file-to-merge [ Review the merged patch, make sure it's ok, build/test/etc. on my box ] % cvs commit file-to-merge

This is alot of work, but it *does* the job. Whole-scale merging of trees and directories doesn't work when the trees have already been merged. CVS doesn't operate on the principal of 'everlasting' branches that keep getting 'merges' happen to them. As I understand the CVS model, if we merged everything into -stable, then we should build a new branch at that point in -current since the files are merged.

Merges are 'merges', not 'sort of merges'.

Nate