[Buildroot] [PATCH 2/2] package/python-terminaltables: fix build backend

[James Hilliard]

On Tue, Jul 11, 2023 at 3:55 AM Thomas Petazzoni
<thomas.petazzoni at bootlin.com> wrote:
>
> On Tue, 11 Jul 2023 03:43:26 -0600
> James Hilliard <james.hilliard1 at gmail.com> wrote:
>
> > This series can be reviewed entirely independently of the other one, I
> > went ahead and marked the final pep517 setuptools migration patch as
> > deferred for now so that there aren't dependency ordering issues or
> > confusion regarding the relation of this series to the other one.
>
> This is a bad idea, which causes even more confusion, as we don't
> understand *why* you're making those changes.

The build system is supposed to be poetry/poetry-core, not setuptools, poetry
is the correct backend specified in the pyproject.toml, the sdist has
a setuptools
shim for backwards compatibility apparently, this shim isn't even checked into
version control for the project so we shouldn't be using it as it's
not really the
correct way to be building a poetry based package.

https://github.com/matthewdeanmartin/terminaltables/blob/v3.1.10/pyproject.toml#L64-L66

>
> Could you instead do what we suggest, i.e resend one single full series
> that include all changes, including the final change of the PEP517
> setuptools migration, together with a cover letter?

I mean, this is just a bug that didn't get noticed earlier, I just hadn't caught
it before I did my pep517 migration, the fix is the same whether or not
the pep517 patch is merged.

If I had caught these bugs before sending the setuptools pep517 migration
series I would have deferred the final migration patch there to avoid this
sort of confusion.

>
> I've already asked this in 3 separate e-mails, but you insist on not
> doing what is requested to get your changes integrated :-/

I've sent a bunch of patches that effectively do the same thing as this one
which have been merged independently, so other than the dependency
ordering I don't see how this is related to the setuptools pep517 migration
patch. To me it seems that I accidentally created a bunch of unnecessary
confusion by sending my setuptools pep517 migration patch too early when
I should have just waited until dependency fixes like this were merged.

I'm kind of pushing back a bit here since making a giant patch series would
just end up making reviewing/rebasing a lot more confusing IMO.

If this was a change that couldn't be justified independently I would agree
that it should be in a series with others.

I think there also might be something wrong with how I'm managing my
patch series with git as they seem to be significantly easier for others to
manage. Every time I deal with a large series it feels like it's a lot harder
than it should be.

>
> Thomas
> --
> Thomas Petazzoni, co-owner and CEO, Bootlin
> Embedded Linux and Kernel engineering and training
> https://bootlin.com