> On Dec 19, 2018, at 12:54 PM, Michael Catanzaro wrote:
>
> On Tue, Dec 18, 2018 at 2:31 PM, Ryosuke Niwa wrote:
>> I tend to agree but then we'd come up with other numbers for the empty &
>> deleted values.
>> I've been thinking that we could use -1 and -2 but that's also somewhat
>> arbitrary restriction.
>
> Using min/max values seems much safer. E.g. we already have in HashTraits.h:
I think that’s true for integers, but for pointers, 0 and -1 are likely safer
(if you don’t need to store null pointers) or -1 and -2 if you do, since all 1
bits is less likely to be a valid address. There’s also some optimization for
the special case where the empty value is 0, but probably not that important if
we often need to store null pointers.
> // Default unsigned traits disallow both 0 and max as keys -- use these
> traits to allow zero and disallow max - 1.
> template struct UnsignedWithZeroKeyHashTraits :
> GenericHashTraits {
> static const bool emptyValueIsZero = false;
> static T emptyValue() { return std::numeric_limits::max(); }
> static void constructDeletedValue(T& slot) { slot =
> std::numeric_limits::max() - 1; }
> static bool isDeletedValue(T value) { return value ==
> std::numeric_limits::max() - 1; }
> };
>
> And:
>
> template struct SignedWithZeroKeyHashTraits :
> GenericHashTraits {
> static const bool emptyValueIsZero = false;
> static T emptyValue() { return std::numeric_limits::min(); }
> static void constructDeletedValue(T& slot) { slot =
> std::numeric_limits::max(); }
> static bool isDeletedValue(T value) { return value ==
> std::numeric_limits::max(); }
> };
>
> Which both seem much safer than the current default:
>
> // Default integer traits disallow both 0 and -1 as keys (max value instead
> of -1 for unsigned).
> template struct GenericHashTraitsBase :
> GenericHashTraitsBase {
> static const bool emptyValueIsZero = true;
> static void constructDeletedValue(T& slot) { slot = static_cast(-1); }
> static bool isDeletedValue(T value) { return value == static_cast(-1); }
> };
>
> Michael
>
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev
