On Tue, Sep 13, 2022 at 12:08:48PM +0000, Juraj Linkeš wrote:
>
>
> > -----Original Message-----
> > From: Bruce Richardson <bruce.richard...@intel.com>
> > Sent: Monday, September 12, 2022 5:15 PM
> > To: Owen Hilyard <ohily...@iol.unh.edu>
> > Cc: Juraj Linkeš <juraj.lin...@pantheon.tech>; tho...@monjalon.net;
> > david.march...@redhat.com; ronan.rand...@intel.com;
> > honnappa.nagaraha...@arm.com; lijuan...@intel.com; dev@dpdk.org
> > Subject: Re: [PATCH v4 1/9] dts: add project tools config
> >
> > On Mon, Sep 12, 2022 at 10:06:03AM -0400, Owen Hilyard wrote:
> > > > E203 - whitespace before ‘,’, ‘;’, or ‘:’
> > > > E266 - too many leading ‘#’ for block comment
> > > > E501 - line too long
> > > > E731 - do not assign a lambda expression, use a def
> > > > C0111 - Missing %s docstring
> > > > F0401 - Unable to import %s
> > > E203, E266 and E501 were disabled due to pylama fighting with the
> > > autoformatters, so I decided to let the autoformatters win.
> >
> > That sounds strange. Is there no way to configure the auto-formatters to
> > avoid
> > these issues? E203 I think we definitely should look to re-enable in DTS.
> >
> > > I think
> > > that C0111 was suppressed because this set of suppressions was from
> > > mainline DTS and that has a lot of functions without
> > > documentation.
> >
> > Is this something we can fix as we go, migrating to DPDK main tree?
> >
> > > F0401 is disabled due to dependencies on TRex vendored
> > > python libraries, since those will not be possible to import inside of
> > > the container.
> >
> > That seems fair enough to keep disabled.
> >
> > > I don't remember why E731 is set, but it may be due to
> > > the rte flow rule generator I wrote for mainline DTS, which makes use
> > > of lambdas extensively to enable lazy evaluation, so that DTS doesn't
> > > need to keep ~2 billion rules in memory.
> >
> > That sounds reasonable too.
> >
> > Can we perhaps target getting from 6 disabled warnings to 2 or 3 at most?
> >
> > Other opinions?
> >
>
> We don't actually need the ignorelist at this point - I've removed the
> ignorelist, fixed a few issues and the tools don't report any problems. I'd
> like to start here and disable checks (either globally or in code where it's
> needed) when we actually need them (in a future patch).
>
Great idea. Thanks.
