Add Boolean method to interpret byte value as boolean?

[John Rose]

C treats all non-zero values as true, x != 0.

When it must, the JVM masks off all but the LSB and checks that, (x & 1) == 1.
(See javadoc for MethodHandles#explicitCastArguments for an example;
the JNI wrappers also do this.)

Those are the relevant precedents, not x == 1.
Probably the C one (x != 0) is less surprising for these use cases.

— John

> On Feb 20, 2020, at 9:19 AM, Maurizio Cimadamore <maurizio.cimadamore at oracle.com> wrote:
> 
> 
> On 20/02/2020 16:02, Ty Young wrote:
>> 
>> On 2/20/20 8:13 AM, Maurizio Cimadamore wrote:
>>> Yes - I think 0 == false has enough mathematical backing which would make sense to add it, IMHO. Other APIs do it too [1].
>> 
>> 
>> Would 0 == false be a safe implementation of such a method? Hypothetically if whatever command line switch was passed that allows "dirty" memory, I imagine that this could cause issues since dirty memory could be interpreted as true. Boolean.valueOf(<String>) also checks for true instead of false.
> 
> I guess what I meant was that, from a mathematical perspective, there's enough ground for considering false and 0 as isomorphic. While it would be theoretically possible to imagine a language in which true is encoded as 0 and != 0 means false, that wouldn't be a very interesting language (e.g. because, in such a language, true | false != true)
> 
> So, after we buy that valueOf(0) has to be false, there's the separate secondary question you mention about what should happen when you provide a value that is != 0. Here I don't think there's any clear cut answer - and I'm afraid that just assuming e.g. 1 == true could be overly restrictive in certain cases (although it has the nice property that, if we went down this path, you could also allow for a unique reverse mapping, from boolean back to int).
> 
> Maurizio
> 
>> 
>> 
>>> 
>>> Also, if we had such a method, we could use it to adapt a byte returning VarHandle/MethodHandle into a boolean returning one.
>> 
>> 
>> That would be nice!
>> 
>> 
>>> 
>>> Maurizio
>>> 
>>> [1] - https://commons.apache.org/proper/commons-lang/javadocs/api-2.4/org/apache/commons/lang/BooleanUtils.html#toBoolean(int)
>>> 
>>> On 20/02/2020 13:56, Ty Young wrote:
>>>> Hi,
>>>> 
>>>> 
>>>> Many functions use boolean values to determine whether or not an operation succeeded(true) or failed(false). Project Panama does not actually see booleans however and instead interprets them as byte values(which is technically correct). This means that in order to convey the correct type as defined in C, you would need to do your own check to see if the value is 1.
>>>> 
>>>> 
>>>> There isn't, however, a standard method of doing this. Boolean only has a function for getting a boolean from a string, not a byte. Given that the need to convert a byte value to a boolean will be more common with Project Panama it seems appropriate that one be added.
>>>> 
>>>> 
>>>> Could this be done?