Re: [Gnu-arch-users] "tla commit" generates a patch-set even if there are no changes

[Charles Duffy]

On Fri, 2004-05-14 at 04:56, Julian T. J. Midgley wrote:
> There are good reasons for having sensible default behaviours in a
> product.  Since most people, most of the time, will not wish to commit
> an empty patchlog, and the addition of an empty patchlog, harmless
> though it is, cannot be reversed without doing the unthinkable and
> playing with history, it makes every sense for tla commit to refuse to
> commit an empty patchlog by default.

Ahh, but we're not discussing an empty patchlog -- we're discussing a
non-empty patchlog with a null changeset attached.

This means that someone went and created a non-empty patchlog, and told
tla to commit it. I think it's pretty damn reasonable for tla to
actually commit in this case.

(Unless, of course, "commit -s" was used. Still, this implies that the
user *knows* what changes there are -- they have to be passing something
to the -s, right? -- and so if they've indicated to the software that
they know what their changes are via either commit's -s flag or creating
a log entry, what business is it of tla's to argue?)