I updated the stable branch, but I made a mistake while pushing the
changes. I was trying to push just one changeset to test if things work
correctly. I did not mention the revision id for the changeset that I
wanted to push. Hence now gem5-stable repo is at the same point as gem5
repo.
I am not aware of any way for rolling back to some particular point in
past. Should we leave it as is? Or can we do something about it?
This is exactly what happened:
-------------------------------------------------------------------------
[nilay@ribera] (10)$ hg update -r f58f93f1656c
1498 files updated, 0 files merged, 177 files removed, 0 files unresolved
[nilay@ribera] (11)$ hg tip
changeset: 9756:0b4a08751b42
tag: tip
user: Nilay Vaish <[email protected]>
date: Thu Jun 13 07:24:25 2013 -0500
summary: config: Do not instantiate membus when using ruby
[nilay@ribera] (12)$ hg log | less
[nilay@ribera] (13)$ hg summary
parent: 9074:f58f93f1656c
ARM: Fix issue with predicted next pc being wrong because of advance()
ordering.
branch: default
commit: 1042 unknown (clean)
update: 682 new changesets (update)
mq: 38 unapplied
[nilay@ribera] (14)$
[nilay@ribera] (14)$
[nilay@ribera] (14)$
[nilay@ribera] (14)$
[nilay@ribera] (14)$ hg push ssh://[email protected]/gem5-stable
pushing to ssh://[email protected]/gem5-stable
The authenticity of host 'repo.gem5.org (141.212.106.70)' can't be
established.
RSA key fingerprint is f8:fc:93:39:96:19:22:cd:df:71:ab:24:8d:46:41:74.
Are you sure you want to continue connecting (yes/no)? yes
Enter passphrase for key '/u/n/i/nilay/.ssh/id_rsa':
searching for changes
remote: Warning: Permanently added 'repo.gem5.org' (RSA) to the list of
known hosts.
remote: adding changesets
remote: adding manifests
remote: adding file changes
remote: added 683 changesets with 6830 changes to 1633 files
remote: notify: sending 1 subscribers 683 changes
----------------------------------------------------------------------------
--
Nilay
On Fri, 14 Jun 2013, Nilay Vaish wrote:
So I am going to update the gem5-stable repo over the weekend to the
following revision:
changeset: 9644:07352f119e48
user: Ali Saidi <[email protected]>
date: Mon Apr 22 13:20:31 2013 -0400
summary: cpu: fix a switching issue with the o3 cpu.
If we see any major bugs being reported by users of gem5-stable, I'll try to
see if any patches committed after 07352f119e48 can resolve those and push
them to the stable repo.
--
Nilay
On Fri, 24 May 2013, Steve Reinhardt wrote:
Just to add my two cents... as Ali said, Nilay's points are true but he is
not the first to raise them; we have had multiple lengthy discussions on
this list on defining an appropriate policy for regularly updating
gem5-stable, only to have them all fall apart when it came to following
through.
I'll differ semantically with Nilay and Andreas on a few points though.
First, gem5-stable is stable in the sense that it changes much less
frequently than gem5 (much much much less... too much less... but it does
fit the definition of stable). Second, I'd say that gem5-stable is not a
"separate repository", it really is just a specific revision of the gem5
tree. It's another clone of the same repo as gem5 (and is updated just by
pulling from gem5 as Ali pointed out). It only looks like a separate
repository given the way we expose it on the web site.
I'm all for giving it another go... I don't think we even need to invent a
new policy, just pull out the one we agreed on last time, but find a way to
make it happen. If we conclude that we're incapable of implementing a
policy (a reasonable conclusion based on history) and it's better to get
rid of it than to leave it to get so stale, I'm OK with that too.
Steve
On Thu, May 23, 2013 at 8:56 PM, Ali Saidi <[email protected]> wrote:
You should be able to just push to it.
Something like
hg clone http://repo.gem5.org/gem5
cd gem5
hg update -r <rev that is good/stable/whatever>
hg push ssh://[email protected]/gem5-stable
Ali
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
