Chris >True. Now we could probably fit the toy problems we were able to deal with >inside chip cache, even with Byte boolean:-) But of course those were toy >problems because we could not deal with real ones. Also we did not have the >plethora of HUGE real data that is easily gotten by the likes of Facebook. The >problem has grown a lot faster than even Moores law accommodates. Social >networkers like me could not even dream of the data (in the 70's) we have now.
>Most modern social network stuff is on a single type of relation. Whereas our >methodology dealt transparently with multitypes by looking at the algebraic >relations generated. >K is fantastic for dealing with ragged array issues, another beef i have with >J, but i do not see it as being any better in dealing with boolean arrays. greg ~krsnadas.org -- from: chris burke <[email protected]> to: Programming forum <[email protected]> date: 13 August 2015 at 20:48 subject: Re: [Jprogramming] Bitwise operations utility >True, but I wonder how much of a disadvantage is that in practice? >Consider that kdb+, the language most optimized for huge data sizes, also has >byte booleans. >I suspect that bit booleans were more important in the old days of limited >RAM, and less important now. -- from: greg heil <[email protected]> to: Programming forum <[email protected]> date: 13 August 2015 at 20:22 subject: Re: [Jprogramming] Bitwise operations utility Chris >Just going on the obvious seat of pants issue of (at least, assuming 8 = byte >size) 8x smaller arrays and 8x as much data that need sloshing around through >all the memory hierarchy. logically everything is well representable ... it is >just a problem at a practical level. greg ~krsnadas.org -- from: chris burke <[email protected]> to: Programming forum <[email protected]> date: 13 August 2015 at 19:59 subject: Re: [Jprogramming] Bitwise operations utility >> ...there has been a long standing resistance to having such. >I think the "resistance to having such" is more a matter of not having the >time to do it properly. >> ... is definitely the poorer for those who would use such. >In what way would you use bit booleans that would not be possible with byte >booleans? ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
