Re: git-commit: allow From: line to be entered in commit message
- From
Junio C Hamano <junkio@cox.net>
- Date
- Jan 13, 2006, 20:46 UTC
- Message-ID
- <7vvewn3lgg.fsf@assigned-by-dhcp.cox.net>
- In-Reply-To
- <7virsn50lv.fsf@assigned-by-dhcp.cox.net>
Junio C Hamano <junkio@cox.net> writes:
Show 10 quoted lines
> Joel Becker <Joel.Becker@oracle.com> writes: > >> On Fri, Jan 13, 2006 at 11:39:17AM -0800, Junio C Hamano wrote: >>> That's what I do. Although I use git-am not git-applymbox, both >>> of them are designed to work that way. >> >> While I can see that git-am and git-applymbox have different >> options for the same basic task, I can't quite see why one would be >> preferred to the other. What does git-am do that git-applymbox does >> not?
The behaviour upon seeing unapplicable patch is somewhat different. In Linus workflow, he reviews (and modifies if necessary) all patches inside mbox and runs "applymbox"; upon failure, he blows what remains in .dotest away, trims mbox to get rid of what has already been applied and re-runs it from scratch. The failure recovery method "applymbox" had (this happened before my time IIRC) is to edit .dotest/patch to make it applicable and re-run it. OTOH, "am" tries to do better by allowing you to hand tweak the working tree to match what would have resulted if the patch applied cleanly and say "--resolved".
Another difference is that "am" can be told to handle binary file changes and apply such as long as the patch is intra repository (i.e. both pre and post image blob are available in the repository). This is used as a backend to do "git rebase".