Re: [PATCH v3] commit-tree: add missing --gpg-sign flag

["Martin Ågren" <martin.agren@gmail.com>]

Hi Brandon,

On Sat, 19 Jan 2019 at 19:05, Brandon Richardson
<brandon1024.br@gmail.com> wrote:
> > I looked into this test in a bit more detail, and it seems to be quite
> > hard to get right. Part of the reason is that `git commit-tree` requires
> > a bit more careful use than `git commit`, but part of it is that the
> > tests that we already have for `git commit-tree [-S]` right before the
> > ones you're adding are a bit too loose, IMHO. So they're not ideal for
> > copy-pasting... I've come up with the patch below, which you might want
> > to use as a basis for your work.

> Just finished adding in the changes you suggested, and everything looks
> good on my end. I based my changes on the patch you provided.
>
> > Or, a bit simpler:
> >
> >   oid=$(echo 10 | git commit-tree -S HEAD^{tree}) &&
> >   git tag tenth-signed "$oid"
>
> Just noticed your latest email. Do you prefer it this way?

I think so, yeah. (But who knows what others might prefer? ;-) )

The use of "" around $oid is perhaps a bit subtle, but not too much so,
I think. The "test_line_count" version was probably a bit too paranoid
and verbose, for no real gain.

> If so, I can amend
> what I have before I submit v4.
>
> When I submit v4, should I submit the patch you created as well, given
> that my changes are based off of it?

I think the cleanest would be to submit a two-patch series, v4.

Alternatively, you could submit only a patch of your own, but it should
then be based directly off of origin/master. So the test in it could
be inspired by my patch, but yours would not have mine as a parent and
the context lines of your patch would look like what is currently in
master. My patch could then go on top of yours, as a "the new tests are
more robust than these old ones; let's rewrite them to the new style".

Thanks
Martin