[Gnu-arch-users] Re: BUG: extra diff arguments for "tla changes --diffs"]

[Stefan Monnier]

> An increased number of context lines can lead to more rejects, because the
> extreme context boundaries may not match.

And that's good if you want to avoid bad patching.

> If standard changesets can be generated using whitespace control, the nature
> of changesets changes.  They're no longer "the difference between this tree
> and that tree" but "the difference between this tree and this imaginary tree
> based on that tree".  I think it's worth thinking twice about that.

It doesn't matter for cherry picking.
But yes, it's clearly a bad idea for archives where the changesets need to
be able to generate the exact same files.  But were's talking about
`tla changes' i.e. changesets that are not going to be directly added to
an archive.

> How is applying a changeset generated with -p "usefully different"?
> That's exactly the kind of output-format-altering change that has no benefit
> for applying changesets, but consumes programming and testing resources.

Don't know about you, but I do a fair bit of manual application of
changesets (or at least their rejected parts), in which case things like
`-p' could be convenient.  Admittedly, `patch' throws away this info when
generating the .rej files, (and the info is similarly lost with diff3) so
it's not like passing "-p" is all there is to it.


        Stefan