Re: C99 Initialisers

[Ian Hastie]

On Tuesday 12 Aug 2003 12:27, Matthew Wilcox wrote:
> On Mon, Aug 11, 2003 at 10:38:27PM -0700, Greg KH wrote:
> > On Tue, Aug 12, 2003 at 03:39:36AM +0100, Matthew Wilcox wrote:
> > > I don't think anyone would appreciate you converting that to:
> > >
> > > static struct pci_device_id tg3_pci_tbl[] __devinitdata = {
> > > 	{
> > > 		.vendor		= PCI_VENDOR_ID_BROADCOM,
> > > 		.device		= PCI_DEVICE_ID_TIGON3_5700,
> > > 		.subvendor	= PCI_ANY_ID,
> > > 		.subdevice	= PCI_ANY_ID,
> > > 		.class		= 0,
> > > 		.class_mask	= 0,
> > > 		.driver_data	= 0,
> > > 	},
> >
> > I sure would.  Oh, you can drop the .class, .class_mask, and
> > .driver_data lines, and then it even looks cleaner.
> >
> > I would love to see that kind of change made for pci drivers.
>
> I really strongly disagree.  For a struct that's as well-known as
> pci_device_id, the argument of making it clearer doesn't make sense.

Which is OK for those very familiar with the code, but not so good for anyone 
else.

> It's also not subject to change, unlike say file_operations, so the
> argument of adding new elements without breaking anything is also not
> relevant.

Even if it is not subject to change, which can never be a total certainty, it 
will need to be applied in new code.  The greater clarity provided by the 
C-99 format will make it easier to ensure the appropriate values are put into 
the right fields.

> In tg3, the table definition is already 32 lines long with 2 lines per
> device_id.  In the scheme you favour so much, that becomes 92 lines, for
> no real gain that I can see.

Smaller source code is not everything.  A consistant style is also very 
important.

-- 
Ian.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/