Sign in to follow this  
MARS_999

Bit mask math

Recommended Posts

MARS_999    1627

How can I combine this into one statement? 

 

I tried this but no luck

if(id &= ~(NX::COLLISION_ID_IsPickable | NX::COLLISION_ID_IsHighlightable) == 200)

//do something 

 

 

irr::s32 id = selectedNode->getID();
if(id &= ~NX::COLLISION_ID_IsPickable)
if(id &= ~NX::COLLISION_ID_IsHighlightable)
if(id == 200)
{
//do something
}

Share this post


Link to post
Share on other sites
iMalc    2466

Surely this is more readable?!:

id &= ~(NX::COLLISION_ID_IsPickable | NX::COLLISION_ID_IsHighlightable);
if (id == 200)
Edited by iMalc

Share this post


Link to post
Share on other sites
Krohm    5030

To elaborate on Sean's, I have to admit I have bought myself quite some peace of mind since I've restricted bitmask usage to serialization only. Just use structs.

auto props = selectedNode->getFlags();
if(props.canPick && props.canHilight) {
//do something
}

The profiler will tell you if code is hot or cold. Cold code has no point in not having maintainance and readability as priorities. Even if the code is hot, I would be careful in considering this an optimization.

Share this post


Link to post
Share on other sites
Scarabus2    560

C++ wizardry like that drives me nuts, and is a clear sign of an inexperienced programmer. SeanMiddleditch is totally right.
Clarity always wins over a couple of cycles worth of optimization, especially in a team environment where you aren't the only one dealing with the code. Basically, if you're trying to reduce line counts for arbitrary reasons, then you're doing it wrong. Bit masking and bit shifting is almost always a sign of over-complication.
 

Edited by Scarabus2

Share this post


Link to post
Share on other sites
alvaro    21246


I've programmed C++ for nearly 10 years and I have no idea what "id &= ~NX::COLLISION_ID_IsPickable" even does. (I do, but I'd have to look it up to be sure.)

 

Well, you must not have done a whole lot of bit manipulations in your 10 years of C++. In any case, a helper function can add clarity to the code, so I would do something like this:

 

void clear_bits(irr::s32 &x, irr::s32 bits_to_erase) {
  x &= ~bits_to_erase;
}
 
//...
clear_bits(id, NX::COLLISION_ID_IsPickable);

Share this post


Link to post
Share on other sites
Ameise    1148

Bit fields might also help in this situation, since you can retain the packed nature of having bit-based data but instead accessing the data as C++ struct members for clarity.

Share this post


Link to post
Share on other sites
Trienco    2555

Bit fields might also help in this situation, since you can retain the packed nature of having bit-based data but instead accessing the data as C++ struct members for clarity.

 

One annoying detail to keep in mind: the order in which bits are assigned is not defined. Not an issue if everything is happening on one machine and within the same application.

 

Still, not knowing if setting a particular field will effectively result in 0x01 or in 0x80 is always making me nervous.

Share this post


Link to post
Share on other sites
MARS_999    1627
Álvaro nice function. Like that...

 

The 200 isn't a permanent thing, it was more of a hurry up and post it!

 

I am using irrlicht and the id is a unsigned int and they are bit masking with the pipe | to do collision detection. So I am looking for ways to set and clear and change them without much headache...

 

Thanks!

Share this post


Link to post
Share on other sites
osmanb    2082

Although I share the concerns that everyone else has about the readability and maintainability of this code (I'd stick with something clear over something short)... it's also worth pointing out that pretty much every alternate form that people have posted does not have the same effect as the original code. If you do anything with id after the code that you posted, then the final value of id varies, depending on whether or not it had the Pickable flag set. (Highlighted is only cleared if id had any bit other than pickable). In fact, I'm beginning to seriously doubt that this code does what you think it does. What were you actually trying to do with this logic?

Share this post


Link to post
Share on other sites
Brother Bob    10344

Although I share the concerns that everyone else has about the readability and maintainability of this code (I'd stick with something clear over something short)... it's also worth pointing out that pretty much every alternate form that people have posted does not have the same effect as the original code. If you do anything with id after the code that you posted, then the final value of id varies, depending on whether or not it had the Pickable flag set. (Highlighted is only cleared if id had any bit other than pickable). In fact, I'm beginning to seriously doubt that this code does what you think it does. What were you actually trying to do with this logic?

I actually thought so too that the solutions are different, until I realized that the if-statements clears a bit but checks the entire value and not just that cleared bit. The only way to break on the first of the three if statements is if no bits at all are set in the original value. You are right that highlighted is only cleared if something other than pickable is set, but if nothing else is set then highlighted cannot be set either so there's nothing to clear in the first place (nor is there anything that can be equal to 200).

 

Whether you clear them all at the same time, as MARS and iMalc suggested in their solutions, or one at a time as MARS did in his original code actually makes no difference. All codes effectively clear both bits and checks the remaining value for 200. The net result is the same in all cases, and that includes whether the last if-statement (=200) is take or not, and what the final value of id is.

 

In Krohm's variant though it makes a difference since it doesn't modify anything at all. And, of course, like most other here I prefer a non-modifying variant.

Edited by Brother Bob

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this