threads / discuss / 27130
Subject: rebasing branch with reverted patches
Hello,
I'm wondering if it would be a good idea for git rebase to allow not rebasing reverted patch which are part of the rebased branch.
For example I'm currently rebasing my branch 'devel' onto master. This branch have several commits and specially one called A and another one called A' which reverts A.
When rebasing 'devel' branch, rebase could try to drop both A and A'.
What do you think ?
BTW is there a way to do this currently ?
Thanks
-- Francis
Michael J GruberRe: rebasing branch with reverted patches
Francis Moreau venit, vidit, dixit 19.04.2011 09:32:
> Hello, > > I'm wondering if it would be a good idea for git rebase to allow not > rebasing reverted patch which are part of the rebased branch. > > For example I'm currently rebasing my branch 'devel' onto master. This > branch have several commits and specially one called A and another one > called A' which reverts A. > > When rebasing 'devel' branch, rebase could try to drop both A and A'. > > What do you think ? > > BTW is there a way to do this currently ?
You can do this with "rebase -i" by removing A and A' from the commit list (or squashing them or dealing with them in whatever way you like).
Michael
Re: rebasing branch with reverted patches
On Tue, Apr 19, 2011 at 10:42 AM, Michael J Gruber <git@drmicha.warpmail.net> wrote:
> Francis Moreau venit, vidit, dixit 19.04.2011 09:32: >> Hello, >> >> I'm wondering if it would be a good idea for git rebase to allow not >> rebasing reverted patch which are part of the rebased branch. >> >> For example I'm currently rebasing my branch 'devel' onto master. This >> branch have several commits and specially one called A and another one >> called A' which reverts A. >> >> When rebasing 'devel' branch, rebase could try to drop both A and A'. >> >> What do you think ? >> >> BTW is there a way to do this currently ? > > You can do this with "rebase -i" by removing A and A' from the commit > list (or squashing them or dealing with them in whatever way you like).
Of course, but my point was to make this automatically...
-- Francis
Andreas EricssonRe: rebasing branch with reverted patches
On 04/19/2011 11:09 AM, Francis Moreau wrote:
> On Tue, Apr 19, 2011 at 10:42 AM, Michael J Gruber > <git@drmicha.warpmail.net> wrote: >> Francis Moreau venit, vidit, dixit 19.04.2011 09:32: >>> Hello, >>> >>> I'm wondering if it would be a good idea for git rebase to allow not >>> rebasing reverted patch which are part of the rebased branch. >>> >>> For example I'm currently rebasing my branch 'devel' onto master. This >>> branch have several commits and specially one called A and another one >>> called A' which reverts A. >>> >>> When rebasing 'devel' branch, rebase could try to drop both A and A'. >>> >>> What do you think ? >>> >>> BTW is there a way to do this currently ? >> >> You can do this with "rebase -i" by removing A and A' from the commit >> list (or squashing them or dealing with them in whatever way you like). > > Of course, but my point was to make this automatically... >
That would be akin to removing in-code comments of why it's a bad idea to implement a particular solution in a particular way, and it also removes the capability of reverting the revert (ie, re-doing the change) at a later time when stability can be sacrificed temporarily.
-- Andreas Ericsson andreas.ericsson@op5.se OP5 AB www.op5.se Tel: +46 8-230225 Fax: +46 8-230231 Considering the successes of the wars on alcohol, poverty, drugs and terror, I think we should give some serious thought to declaring war on peace.
Re: rebasing branch with reverted patches
On Tue, Apr 19, 2011 at 11:15 AM, Andreas Ericsson <ae@op5.se> wrote:
> > That would be akin to removing in-code comments of why it's a bad idea > to implement a particular solution in a particular way, and it also > removes the capability of reverting the revert (ie, re-doing the change) > at a later time when stability can be sacrificed temporarily.
That's right but OTOH I use git-rebase for development stuff where all correct points you raised don't apply.
-- Francis
Re: rebasing branch with reverted patches
On Tue, Apr 19, 2011 at 11:09 AM, Francis Moreau <francis.moro@gmail.com> wrote:
> On Tue, Apr 19, 2011 at 10:42 AM, Michael J Gruber > <git@drmicha.warpmail.net> wrote: >> Francis Moreau venit, vidit, dixit 19.04.2011 09:32: >>> Hello, >>> >>> I'm wondering if it would be a good idea for git rebase to allow not >>> rebasing reverted patch which are part of the rebased branch. >>> >>> For example I'm currently rebasing my branch 'devel' onto master. This >>> branch have several commits and specially one called A and another one >>> called A' which reverts A. >>> >>> When rebasing 'devel' branch, rebase could try to drop both A and A'. >>> >>> What do you think ? >>> >>> BTW is there a way to do this currently ? >> >> You can do this with "rebase -i" by removing A and A' from the commit >> list (or squashing them or dealing with them in whatever way you like). > > Of course, but my point was to make this automatically...
You can also change the A' commit title to "fixup! original_title" while reverting (there is no --fixup argument like "git commit" has), then git rebase -i would reorder the todo file for you.
git revert gets confused by the resulting empty change but you can "git rebase --continue", or now that both are next to the other just remove them.
HTH, Santi
Re: rebasing branch with reverted patches
Hello,
On Tue, Apr 19, 2011 at 11:39 AM, Santi Béjar <santi@agolina.net> wrote:
> > You can also change the A' commit title to "fixup! original_title" > while reverting (there is no --fixup argument like "git commit" has), > then git rebase -i would reorder the todo file for you.
Interesting I wasn't aware of --autosquash option, thanks.
This is almost what I was asking for actually, the only part which is missing is the fact that git-rebase detect the "Revert: " magic string, check that the revert is sane (by comparing the diff of the revert with the diff of the original commit) and do the same as "fixup!" would do without the confusion of the resulting empty commit.
Thanks
-- Francis