IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2014/11/07)20141108 
avih tor, Math.random does this: js_pushnumber(J, (double)rand() / (RAND_MAX - 1));, but this can return 1 or even bigger than 1. you probably wanted .. / (RAND_MAX + 1)11:19.10 
  or, if you can't know if int/long/whatever can hold RAND_MAX+1 without overflowing, then just retry if it results in 111:21.22 
  and if you get 1 3 times then it must be a sign from god that something is wrong.11:28.43 
nsz yes that should be RAND_MAX+111:36.55 
  i'd use RAND_MAX+1.0 and then no cast is needed :P11:37.25 
  avih: RAND_MAX+1 is not representible as int usually11:41.13 
  so it should be either RAND_MAX+1U11:41.36 
  or +1.011:41.49 
avih nsz: yeah, 1.0 is nice enough :)13:23.37 
  personally though, when i'm using rand in testing i need it seeded, but js doesn't provide seed api. so i roll my own seeded rand.13:25.08 
  re 1I though, in some definitions it says that RAND_MAX is an integer, at which case 1U would work, but in others i saw that "it's an intergal", at which case it might be itself max uint13:27.48 
  re 1U*13:27.57 
nsz where did you see it as an 'integral'?14:02.01 
  the c standard is pretty clear14:02.09 
  rand() returns int14:02.15 
  and randmax is the maximal returned value14:02.30 
  so RAND_MAX cannot be larger than INT_MAX14:02.41 
  and usually it is INT_MAX (that's the most useful RAND_MAX)14:02.56 
  the type of RAND_MAX may be different but that does not matter14:03.25 
avih http://www.cplusplus.com/reference/cstdlib/RAND_MAX/16:17.43 
  - cstdlib16:18.01 
nsz yes random websites are usually wrong about the c standard unfortunately17:06.59 
  there is no such thing as 'integral constant expression' the corret terminology is 'integer constant expression' but it has not relevance here17:07.44 
  the only relevant part is that it's the maximum value rand may return17:08.06 
  which is at most INT_MAX17:08.11 
 Forward 1 day (to 2014/11/09)>>> 
ghostscript.com
Search: