Messages in this thread | | | Date | Thu, 9 Apr 2009 11:26:49 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: [PATCH 2/2] FRV: Move to arch/frv/include/asm/ |
| |
On Thu, 9 Apr 2009, Al Viro wrote: > > We can teach patch(1) to handle those - apparently agruen has resurrected > development lately (git://git.savannah.gnu.org/patch.git), so it might > be possible without usual latency problems...
I think a lot of SCM's parse diffs internally, so people are still going to have to live with legacy patches.
The reason for doing diff parsing internally is that it's much easier to handle the error cases, or even just see what files were modified successfully.
If you executee an external "patch" command, it's really quite nasty to parse the whole "what happened" stuff both for the "everything applied fine" _and_ for the "oops, the patch failed in the middle" case.
Doing diff parsing and application is also generally quite easy - at least compared to _generating_ them. In many ways it's easier to just handle the diff yourself than it is to try to parse the end result of running "patch" externally.
So I would not be at all surprised if many SCM's use an external "diff" to generate patches (that has none of the complexities of the error cases), but don't use an external "patch" to apply them.
Linus
| |