Messages in this thread | | | Date | Mon, 24 Aug 2009 13:20:40 +1000 | From | Paul Mackerras <> | Subject | Re: [tip:timers/core] timekeeping: Increase granularity of read_persistent_clock() |
| |
Ingo Molnar writes:
> Do you ask Linus to rebase the upstream kernel as well, if the > powerpc or x86 build happens to break? There's more than a dozen > such cases per development cycle triggering on my tests alone. If > not, why not?
I see you pulling commits out of the tip tree quite often, when they have testing failures of various kinds. I presume that, like other maintainers, you have some branches that you try hard not to rebase and other testing branches that are quite volatile and get reconstructed frequently (though I don't know what branch names you use for them).
I presumed that you wouldn't have put a commit that hadn't even passed basic build testing into one of your non-rebasing branches. That's why I assumed you could fold the fix into the original patch without difficulty.
> The thing is, we'll probably redo this portion of the timer tree as > i found other problems in testing, but generally the disadvantages > of a build breakage with a very small non-bisectability window has > to be weighed against the disadvantages of a rebase (which are > significant). > > The equation does not automatically flip in favor of a rebase as you > seem to suggest - in fact it generally goes _against_ a rebase.
In a stable, non-rebasing branch, sure. But putting untested patches into such a branch would be a bit silly, so I assumed you hadn't done that. :)
Paul.
| |