Re: [PATCH (resend)] Pass -C1 to git-apply in StGIT's apply_diff() and apply_patch().
- From
- Catalin Marinas <catalin.marinas@gmail.com>
- Date
- Apr 10, 2007, 22:38 UTC
- Message-ID
- <b0943d9e0704101538p3de0bf56m7906cfe2f5fc157e@mail.gmail.com>
- In-Reply-To
- <20070410193214.GF4946@moonlight.home>
On 10/04/07, Tomash Brechko <tomash.brechko@gmail.com> wrote:
Show 6 quoted lines
> Once we are talking about StGIT's push (push of > the patch back to the stack), why would we want to start tree-way > merge when the context has changed? My point was exactly that since I > want to keep my patches up-to-date with the main branch, I do rebase > from time to time, and I'm not interested in doing the merge every > time just because something has changed upstream in surrounding code.
When something has changed in the surrounding code (not touched by your patch), the automatic three-way merge should, in general, be able to solve the issue as it uses the ancestor information. Is the automatic three-way merge failing as well in your case?
> The same goes for patches that were already applied upstream. > Whatever the current context around the code of my applied patch is, I > have to accept it, because the patch was applied. I'm going to throw > it away locally, but currently I have to do the merge first.
I think -C1 should be OK for merge detection (in most situations) and importing patch files (via import, fold) but I personally don't like it when rebasing a patch. I still prefer a more precise context checking, rather than the fuzzy one similar to the "patch" tool (as the line numbers are usually volatile).
I'm OK with the idea of this patch but I would prefer a config option and/or command line option rather than hard-coding it for people with different views. A command line option could make sense for commands like import/fold and a config option for the rest.
Thanks.
-- Catalin