Reply to: Who'd be a game dev...

[aussielong]

You're damn right I wouldn't, you wouldn't make it past DodgyAgent.

Good comeback!

[SupremeSpod]

Right shifting a negative signed integer is undefined. Or rather , the result is platform and compiler dependent.

You wouldn't be interviewing me pal.

You're damn right I wouldn't, you wouldn't make it past DodgyAgent.

[aussielong]

So, by your logic...


unsigned int a = 4;

a = a >> 1;

The answer could be unpredictable? Sorry mate you failed the interview.

Right shifting a negative signed integer is undefined. Or rather , the result is platform and compiler dependent.

You wouldn't be interviewing me pal.

[SupremeSpod]

It doesn't matter if you are only after the lower bits. It's faster than padding unnecessarily.

Another bloke who does not know his limitations.

So, by your logic...


unsigned int a = 4;

a = a >> 1;

The answer could be unpredictable? Sorry mate you failed the interview.

[aussielong]

I think it's a bit over the line trolling on technical stuff like AL is doing, people might think he's correct and muck up their work!

Get orf moi laaand!!

[d000hg]

I think it's a bit over the line trolling on technical stuff like AL is doing, people might think he's correct and muck up their work!

[chef]

I'm with DA on this one, reading this thread is interesting but also completely over my head. Can someone give a small example of >> << <<< and >>> in some code please so I have a better chance of actually keeping up

[aussielong]

How can you leave the top bits undefined? You'd **** the value up.

Spod - In "I ******* Despair" mode!

It doesn't matter if you are only after the lower bits. It's faster than padding unnecessarily.

Another bloke who does not know his limitations.

[SupremeSpod]

These operators make it fast to multiply and divide by powers of 2, which can be done using others operators, clearly. Some compilers will replace a multiply by 2 with a bit shift to the right.

You get 0 padding if you use >>>, I think with >> you don't... Which leaves the top bits undefined after the shift.

How can you leave the top bits undefined? You'd **** the value up.

Spod - In "I ******* Despair" mode!

[aussielong]

You are wrong.



How do you tell the compiler to do a bit shift, other than using the bitshift operators? And then what does it optimise it to?

Well now you're just making stuff up. What does it pad with if it's not 0? 1?

If it's a signed value, and you shift right, it pads with 0 if positive or 1 if negative. If it's unsigned, it pads with 0. Java seems to have two seperate >> and >>> operators, for some inexplicable reason.

These operators make it fast to multiply and divide by powers of 2, which can be done using others operators, clearly. Some compilers will replace a multiply by 2 with a bit shift to the left.

You get 0 padding if you use >>>, I think with >> you don't... Which leaves the top bits undefined after the shift.

This shouldn't be in general really should it.

[Bunk]

You are wrong.



How do you tell the compiler to do a bit shift, other than using the bitshift operators? And then what does it optimise it to?



Well now you're just making stuff up. What does it pad with if it's not 0? 1?

If it's a signed value, and you shift right, it pads with 0 if positive or 1 if negative. If it's unsigned, it pads with 0. Java seems to have two seperate >> and >>> operators, for some inexplicable reason.

If it's the same as Javascript then the >>> pads it with zero whether it's positive or negative, hence it's the unsigned right shift operator.

[d000hg]

Was ANY part of AL's post accurate

Although a compiler COULD change a multiplication/division into a shift operation if it wanted to, if you are multiplying by a constant e.g x *= 32.

[VectraMan]

>> is from C++, I don't recall seeing it in C but I haven't programmed in C for 15 years so I could be wrong.

You are wrong.

I don't think it's used much because the compiler should do it for you these days anyway and I guess with fast hardware these days, getting the last millisecond performance out of the code is not most peoples concern.

How do you tell the compiler to do a bit shift, other than using the bitshift operators? And then what does it optimise it to?

>>> is C++ too. I think it similar to >> but one pads with 0 and one doesn't. Java is going back to C++ with increasingly complicated and ugly syntax.

Well now you're just making stuff up. What does it pad with if it's not 0? 1?

If it's a signed value, and you shift right, it pads with 0 if positive or 1 if negative. If it's unsigned, it pads with 0. Java seems to have two seperate >> and >>> operators, for some inexplicable reason.

[EternalOptimist]

[DodgyAgent]