On Tue, Mar 24, 2020 at 11:02:00AM +0200, Andy Shevchenko wrote:
> On Tue, Mar 24, 2020 at 09:13:28AM +0100, Wolfram Sang wrote:
> > Hi Andy,
> >
> > > > Here, the initialization to 0 is missing, so some values are broken.
> > >
> > > Yes, and this is fine. They are not being used. So, the idea is, whenever we
> > > pass "false" as a parameter to the function we must take care of all fields we
> > > are using.
> >
> > Can be argued. Still, uninitialized values look a little sloppy IMO. I
> > had a patch on top of this series to print the generated values as debug
> > output, and '0' looks much more intentional there.
> >
> > > > Why don't we just drop the pointer and init the array directly?
> > > >
> > > > struct i2c_timings t = {
> > > > .bus_freq_hz = ...
> > > > ...
> > > > }
> > >
> > > I can do it if you think it's better. I have no strong opinion here.
> > > From code prospective I guess it will be something similar anyway.
> >
> > I like it better. Easier to read in the code, no need for a seperate
> > pointer. I can fix it locally here, though.
>
> I already sent v4 the other day, but can update since I have got new tags to
> pick up.
Okay, v5 is fine with me as well.
Attachment:
signature.asc
Description: PGP signature
