Re: [Qemu-devel] [7234] Use a more natural order

[Blue Swirl]

On 4/23/09, Lennart Sorensen <address@hidden> wrote:
> On Thu, Apr 23, 2009 at 10:01:43PM +0300, Blue Swirl wrote:
>  > True, but it's not the style that is used here. You are of course free
>  > to argue for using this version and even submit patches.
>
>
> I didn't check if it is covered by the codestyle or not for qemu.
>
>  If your patch comment had said "Fix to follow code style" rather than
>  "make natural order" I wouldn't have said anything.  Natural order in
>  this case is a bad reason for the change.  Following code style is a
>  good reason (even if the code style may then be questionable).
>
>  If the code style doesn't cover it at all, I would say the change is
>  horrible and shouldn't be done at all.  The change even caused a compile
>  failure briefly (and hence breaks git bisect) by being commited completely
>  untested I suspect.  If it didn't cause a compile failure it at least
>  should have caused a serious warning.

I don't think any code style document can cover all possible cases.
But another approach can be used: you could try to find a precedent
case where this style has been used in QEMU.

>  > We humans also tend to use base 10 arithmetic and infix notation
>  > despite their well known shortcomings.
>
>
> Well only a small part of humans still use broken date formats.  We do
>  still seem to be atracted to base 10 though.  And of course our time
>  measuring system is just weird.

The formats and systems with varying level of brokenness reflect the
centuries of weird history behind them. Only from a purely engineering
standpoint that is not a valid reason for still using them.