On 01/19/16 18:04, John Snow wrote: > > > On 01/19/2016 06:48 AM, Paolo Bonzini wrote: >> >> >> On 19/01/2016 05:51, John Snow wrote: >>> + /* Only RESET is allowed to an ATAPI device while BSY and/or DRQ are >>> set. */ >>> + if (s->status & (BUSY_STAT|DRQ_STAT)) { >>> + if (!(val == WIN_DEVICE_RESET) && (s->drive_kind == IDE_CD)) { >> >> I was going to complain about Pascal-ish parentheses, but actually I >> think there is a bug here; the expression just looks weird. >> >> Did you mean >> >> if (!(val == WIN_DEVICE_RESET && s->drive_kind == IDE_CD)) >> >> or equivalently applying de Morgan's law: >> >> if (s->drive_kind != IDE_CD || val != WIN_DEVICE_RESET) >> >> ? >> >> Paolo >> >>> + return; >> > > ugh, yes, I typo'd. Thank you. > > If you're still up, which do you find more readable? > The (!(A && B)) form or the (!A || !B) form?
You didn't ask me, but that's no problem for me. :) The logical negation operator "!" has much-much stronger binding than the logical "and" and logical "or" ones. If you use the first form, !(A && B) it works, but spacetime will curl every time someone sees those parentheses overriding the nice n' tight binding of "!". So, for me, only the second form *exists* -- for me, the operand of the logical negation operator must always be as "indivisible" an expression as possible :) So, (!A || !B) without a doubt. Thanks Laszlo