Hacker News Reader

Comment by vasco | Hacker News Reader

I use them a lot when I'm designing hardware, but I never use them when I'm doing "regular programming", there's no reason to do it if I can use more readable, higher level constructs.

View on HN

When writing network protocols or de-marshalling binary data it's common to want to know which bits are set in a byte. You might do a & 4 == 4 to test the 3rd bit is set, how would you do this with 'regular programming'?

I'm guessing that writing networking protocols and even de-marshaling binary data doesn't fall under "regular programming". The vast majority of programming is business programming and marshalling and de-marshalling data is relatively troublesome (Ruby anyone?). Most of these projects are better off using a library than implementing their own.

I'm sad to say I've seen code where the answer is sprintf hex output into a temp string and some string functions to see if the last single char substring eq and/or neq certain values at the string level correlating with the 3rd bit being set (so ends in hex 2, well that means bit 3 can't be set... ends in hex 4 ding ding ding we have a winner) Only a slight upgrade in readability on a chain of mystifying if/then/else is a case statement with 16 possibilities. Only a slight upgrade on that is "optimizing" for speed by squirting out octal so your lookup stable is half the size of a hex one.

    struct header __atribute__((packed)) {
      unsigned int highnibble : 4;
      unsigned int flag_foo : 1;
      unsigned int flag_bar : 1;
      unsigned int flag_baz : 1;
      unsigned int flag_qux : 1;
    }
    
    if (((header)value).flag_bar) {

Reply on news.ycombinator.com