checking bits

Discussion in 'C++' started by Mark, Jan 19, 2006.

  1. Mark

    Mark Guest

    lets say i have a char, and i want to check if the 3rd bit is 0 or 1...
    how might i do this?

    Mark, Jan 19, 2006
    1. Advertisements

  2. Mark

    Mike Wahler Guest

    Learn about the bitwise operators,
    e.g. | , & , etc.

    Also you should used unsigned characters for this.

    Mike Wahler, Jan 19, 2006
    1. Advertisements

  3. Mark

    Mike Wahler Guest

    Alternatively, there's the 'std::bitset' type
    from the standard library.

    Mike Wahler, Jan 19, 2006
  4. Mark

    Mark Guest

    Mark, Jan 19, 2006
  5. Mark

    Shark Guest

    It will help if you want your answer to be the same regardless of
    whether you are on a little endian or big endian machine. What you mean
    by the '3rd' bit can have different meanings.
    Shark, Jan 19, 2006
  6. Mark

    Shark Guest

    There is something about bit manipulations using #define here:

    9. Embedded systems always require the user to manipulate bits in
    registers or variables. Given an integer variable a, write two code
    fragments. The first should set bit 3 of a. The second should clear bit
    3 of a. In both cases, the remaining bits should be unmodified.

    * Use #defines and bit masks. This is a highly portable method and is
    the one that should be used. My optimal solution to this problem would

    #define BIT3 (0x1
    static int a;

    void set_bit3(void) { a |= BIT3; }
    void clear_bit3(void) { a &= ~BIT3; }

    how does this sound!
    Shark, Jan 20, 2006
  7. Whoa!... What's wrong with your newsreading software? And why are
    you writing "0x1" instead of simply "1"? Is there a difference?
    And why not have an argument for those functions? Something like

    void set_bit3(int &a) { a |= BIT3; }

    ? Otherwise you only can use those functions to set or clear the
    bits in some static object named 'a' in this translation unit...

    Victor Bazarov, Jan 20, 2006
  8. Mark

    Shark Guest

    And why not have an argument for those functions? Something like
    Well, I simply copied and pasted the code from . The question was
    aimed at programmers for embedded systems. Maybe they like coding this
    way! But yes, if i were to write this code, I'd pass an argument and
    not use a static int.
    Shark, Jan 20, 2006
  9. Mark

    Jim Langston Guest

    Personally, I've always wanted to be able to do something like:
    #define BIT3 0b00000100
    but 0x is as close as you can get.
    At least it gives the intent. Which would be the same reason he wouldn't do
    #define BIT3 4
    because the intent of 4 is not clear.
    Jim Langston, Jan 22, 2006
  10. Mark

    Mark Guest

    well... the point of defining BIT3 as 4 would be so that it WOULD be
    clear when you used "BIT3".. the definition itself doesn't need to be
    quite so clear does it? anyways "BIT3" should tell you something about
    your intent.

    thanks for the help again guys.
    Mark, Jan 22, 2006
  11. Mark

    Shark Guest

    The meaning of "clear" is not clear here. Do you mean clear as in
    clarity or clear as in "clear the bit"?

    besides. if we #define BIT3 0b00000100 it assumes that we are working
    on a 8-bit number, while in C++ the most basic unit (i could be wrong)
    int is at least 2 bytes. With (0x1<<3) we just don't have to tell what
    size primitives we are working on. At least that is my guess. Again, I
    could be wrong.
    Shark, Jan 22, 2006
  12. Mark

    Luke Meyers Guest

    Portable, sure. Type-safe, no. Maintainable, no. Recommendable,
    absolutely not.

    This is what const is for.

    const unsigned char BIT3 = ...
    So you're starting counting at 0? I love zero-indexing as much as the
    next fella, but that's not a reasonable convention when speaking in
    terms of ordinal numbers. The rightmost bit is the first bit, not the
    zeroth bit. The bit pattern you've produced is 00001000. Surely we
    can all agree that's the *fourth* bit?
    Pretty appalling. Why on earth would someone write a function specific
    to one bit position? I'd vastly prefer something like the following
    (switching back to zero-index mode):

    typedef unsigned char Byte;

    inline Byte const nth_bit(size_t n) { return 0x1 << n; }
    inline Byte const set_bit(Byte src, size_t n) { return src |
    nth_bit(n); }
    inline Byte const clear_bit(Byte src, size_t n) { return src &
    ~nth_bit(n); }

    Luke Meyers, Jan 22, 2006
  13. Mark

    Luke Meyers Guest

    You are wrong. The most basic unit is char. See section 3.9.1 of the
    standard. That said, the size of a char is not guaranteed to be 8
    bits. However, as in mathematics, leading zeros are implicit. That's
    why we can say 0x1 rather than 0x0001 or 0x00000001.

    Luke Meyers, Jan 22, 2006
  14. Mark

    Shark Guest

    Well, once again you are forgetting the context. That question and the
    remark I posted was from an embedded programmer's perspective. They
    worry too much about how many functions are defined where, how much
    memory and crap are assigned.
    Shark, Jan 22, 2006
  15. Mark

    Luke Meyers Guest

    Please don't top-post. I don't know what you mean by "once again," nor
    do I see how anything I said is any less applicable to embedded
    programmers. On the contrary, defining a *single* function to handle
    bit-setting logic rather than a separate function for each individual
    bit is exactly the kind of economy embedded systems programmers are
    concerned with.

    What exactly did you think was wrong about what I said? Would you care
    to comment on the relative technical merits of the two solutions?

    Luke Meyers, Jan 22, 2006
  16. Mark

    Shark Guest

    Oh well, the evils of top posting :p

    I use google to read this ng so I see a nice "thread" of everyone's
    replies and OP and sometimes I forget to scroll all the way down and
    snip this and snip that because the context is a couple of lines above.
    My bad.
    This thread on google is here:

    I made the original quote from this link:

    The particular question 9 reads as follows:

    Bit manipulation
    9. Embedded systems always require the user to manipulate bits in
    registers or variables. Given an integer variable a, write two code
    fragments. The first should set bit 3 of a. The second should clear bit
    3 of a. In both cases, the remaining bits should be unmodified.

    and the solution was:

    #define BIT3 (0x1
    static int a;

    copied verbatim (and source was mentioned in my post). So basically
    I've been corrected a couple of times for something I referenced from
    another site :) I only know that the question was aimed at embedded
    programmers and maybe someone who actually works as one can tell why
    you'd write a macro to set some bit.

    I worked on windows ce devices once and there were some weird things
    going on all over the code, but the terminal was fast enough (400mhz)
    to run badly written mfc apps that leaked memory. There were also some
    hardcore programmer who programmed for single chip barcode scanners and
    tried to squeeze out every single clock cycle they could. Frankly I
    don't know what is the reason behind this #define except that it helps
    skip some symbols that a code-with-functions will put in the compiled
    code. But as always, I could be wrong. I don't have much experience in
    the industry.
    First, I didn't really say you were wrong. I wanted to say that the
    posting I made wasn't really a general solution, but was an answer to a
    question. The question made a specific demand.: write two functions
    that do the following, and the given answer satisfied the requirement.
    So I was only pointing out the context.

    Regarding the relative merits, I'd like to run away from the battle so
    I can fight another day. I just don't have enough experience with
    embedded devices to make a comparison. Maybe someone else can shed a
    light. I hope the meaning of "is" is "clear" now! Peace
    Shark, Jan 23, 2006
  17. Mark

    Luke Meyers Guest

    Ah, no wonder. It's not even a C++ article, it's explicitly a C
    article. How embarassing! ;)
    Well, you referenced it; presumably you thought it was somehow
    I could get on your case for disregarding the *relevant* context (a C++
    newsgroup, the OP's question etc.), but... nah.... ;)
    Werd, and cheers to that. Nice posting with you.

    Luke Meyers, Jan 23, 2006
  18. Mark

    Shark Guest

    Yes it is helpful. That example provided a way to set/unset 3rd bit of
    an int. And I was referencing it because it is useful to the OP's

    If you can explicitly set the 3rd bit of the char's copy, and then
    compare the result with the original then you can determine if the 3rd
    bit is set or not.

    int check3rdbit(char a)
    char a='A';
    char b= set3rdbitof(a); //returns a char with 3rd bit set
    if(a==b) then return THIRD_BIT_IS_SET;
    else return THIRD_BIT_IS_NOT_SET;

    I thought that should be obvious ;) So I wasn't off topic. Plus, if its
    explicitly C, what makes it not C++? after all that page wasn't
    discussing implicit void* casts to char* and like!!!
    haha. "Quien en tiempo huye, en tiempo acude."
    Shark, Jan 23, 2006
  19. Mark

    Luke Meyers Guest

    Yes, but the point is that it's just silly to hard-code '3' into the
    function. What happens when you're interested in bit 4 and bit 7?
    Write two new functions? Or write a single function that can handle
    any of these cases?
    Well, C is not C++. The fact that C++ has a C-like subset does not
    mean that any "good C practice" is a good C++ practice. Implicit void*
    casts to char are one example. Another is the use of #define for
    constants. There's absolutely no reason to do that in C++.

    Nobody said it was off-topic. Just lousy advice.
    Vad sa du?

    Luke Meyers, Jan 23, 2006
  20. Mark

    Shark Guest

    Yes I agree with you. I am against the use of c++ preprocessor to
    define globals (partly because I read Scott Meyers). But the snippet I
    was demonstrating was correct in itself (due to closure, i think.
    Question satisfied by answer, and answer is to the point). Meep!

    If I were to write something like this for my own use, I'd write
    exactly what you suggested. But if I am quoting from another
    source....and if they have a good reason for writing a macro instead of
    function, then ranting is pointless.
    Well, C != C++ obviously, but C is a subset of C++. Just because you
    are using a C++ compiler shouldn't mean you must use all the WMDs that
    C++ provides. Lets use some polymorphism to relate C and C++, maybe the
    most crude way to represent it is this:

    class C++ : public C {
    // insert stroustrupism here
    // override structs, unions, implicit void* conversions.....

    so what does polymorphism tell you? if you derive from a base class,
    there is an "is-a" relationship between the base class and derived
    class. So basically C++ is a type of C. (That is also true for most
    compilers, they convert C++ code to C and then use a C compiler to
    create an executable.)

    Does that mean anywhere you can use C you can also use C++? Sure you
    can! Does that mean C++ is C? Open question. Because years of
    conditioning by books, professors, and crappy recruiters looking for
    "C++ Professionals" has skewed the answer in favor of "no".
    translates into english as "He that fights and runs away, lives to
    fight another day."
    Shark, Jan 23, 2006
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.