2016-07-26T00:21:03Z rumbler31 joined #ccl 2016-07-26T00:30:15Z clop: looking through l0-bignum64.lisp and noticing that you use 32-bit digits -- wondering if there's any quick rationale for why 32 instead of 64 bits? (i don't think it's a bad decision, just looking for wisdom from someone who's thought it over) 2016-07-26T00:36:46Z rme: clop: There are, I think, two reasons. 2016-07-26T00:38:36Z rme: The first is that the existing code used 32-bit bignum digits, and leaving that in place made porting easier (initially to 64-bit PowerPC, and subsequently to 64-bit x86). 2016-07-26T00:39:32Z rme: Secondly, being able to say (uvref b i) is handy becaue then tt's really easy to implement things like minusp for bignums. 2016-07-26T00:41:02Z rme: With a little bit of fiuss, we can (and do) perform arithmetic on bignums 64 bits at a time, even with 32-bit bignum digits. 2016-07-26T00:41:16Z milanj quit (Quit: This computer has gone to sleep) 2016-07-26T00:44:02Z clop: aha thanks 2016-07-26T00:44:51Z clop: i can imagine (perhaps) that 32-bits might occasionally be useful when doing things like multiplies, e.g., you can presumably multiply 2 32-bit digits with a single machine instruction without running into overflow 2016-07-26T00:44:57Z rme: I think it would be fun (I am strange, I know) to implement bignums with 64-bit bignum digits. We could also benefit from using one of the sub-quadratic multiplication algorithms (Karatsuba, Toom-Cook) for bugnum mulitplication, but that's somewhat orthoganal to a 64-bit bignum digit size, really. 2016-07-26T00:44:58Z clop: but i don't know whether that's actually useful 2016-07-26T00:45:41Z clop: right, if i understand correctly those are only useful for very large numbers though, right? 2016-07-26T00:46:41Z rme: IIRC, they can start to win at lengths of 500 bits or so. 2016-07-26T00:47:06Z clop: aah ok thanks 2016-07-26T03:56:25Z rumbler31 quit (Ping timeout: 244 seconds) 2016-07-26T06:19:22Z rme quit (Quit: rme) 2016-07-26T06:47:03Z milanj joined #ccl 2016-07-26T07:10:09Z DavidGu joined #ccl 2016-07-26T07:14:52Z DavidGu quit (Client Quit) 2016-07-26T07:16:06Z DavidGu joined #ccl 2016-07-26T07:54:44Z milanj quit (Quit: This computer has gone to sleep) 2016-07-26T08:07:03Z dfigrish joined #ccl 2016-07-26T08:16:15Z hhdave joined #ccl 2016-07-26T08:17:11Z edgar-rft quit (Quit: edgar-rft) 2016-07-26T08:20:55Z hhdave quit (Ping timeout: 258 seconds) 2016-07-26T09:10:14Z milanj joined #ccl 2016-07-26T09:14:42Z hhdave joined #ccl 2016-07-26T09:55:47Z vydd joined #ccl 2016-07-26T09:55:47Z vydd quit (Changing host) 2016-07-26T09:55:47Z vydd joined #ccl 2016-07-26T10:22:54Z pierpa joined #ccl 2016-07-26T11:23:10Z dfigrish quit (Ping timeout: 252 seconds) 2016-07-26T11:27:04Z Tennis left #ccl 2016-07-26T12:31:42Z schjetne quit (Ping timeout: 260 seconds) 2016-07-26T12:43:53Z mrottenkolber joined #ccl 2016-07-26T12:44:55Z DavidGu quit (Quit: DavidGu) 2016-07-26T13:23:35Z mrottenkolber: when ccl::cfp-lfun returns nil (when called on the first value you get from map-call-frames), what does that mean? 2016-07-26T13:27:39Z mrottenkolber: E.g. what are these (NIL #) stack frames, I guess? 2016-07-26T13:28:50Z mrottenkolber: I am trying to figure out how to interpret them from a profiler perspective. 2016-07-26T13:46:08Z ChibaPet is now known as Landru 2016-07-26T13:56:30Z schjetne joined #ccl 2016-07-26T14:06:27Z DavidGu joined #ccl 2016-07-26T14:20:48Z DavidGu quit (Remote host closed the connection) 2016-07-26T14:21:08Z DavidGu joined #ccl 2016-07-26T14:29:04Z rme joined #ccl 2016-07-26T14:43:19Z pierpa quit (Ping timeout: 260 seconds) 2016-07-26T14:52:08Z mrottenkolber: repeating since rme might now this: when ccl::cfp-lfun returns nil (when called on the first value you get from map-call-frames), what does that mean? 2016-07-26T14:54:06Z mrottenkolber left #ccl 2016-07-26T14:58:02Z mrottenkolber joined #ccl 2016-07-26T15:01:59Z rme: mrottenkolber: I don't remember (if I ever knew) the details, but there can be frames on the stack that don't belong to functions. 2016-07-26T15:03:58Z vydd quit (Remote host closed the connection) 2016-07-26T15:13:48Z mrottenkolber: do they then technically belong to the next upper (or lower) stack frame/function? 2016-07-26T15:21:30Z mrottenkolber: (which is non-NIL)? 2016-07-26T15:52:58Z vydd joined #ccl 2016-07-26T16:05:51Z schjetne quit (Ping timeout: 276 seconds) 2016-07-26T16:22:21Z rme quit (Ping timeout: 183 seconds) 2016-07-26T16:23:04Z rme_ joined #ccl 2016-07-26T16:23:17Z rme_ quit (Client Quit) 2016-07-26T16:24:41Z rme quit (Ping timeout: 258 seconds) 2016-07-26T16:25:24Z mrottenkolber left #ccl 2016-07-26T16:39:24Z hhdave quit (Ping timeout: 260 seconds) 2016-07-26T16:41:53Z Landru is now known as TwistedFruit 2016-07-26T16:50:53Z rme joined #ccl 2016-07-26T17:24:32Z schjetne joined #ccl 2016-07-26T17:53:15Z DavidGu quit (Quit: DavidGu) 2016-07-26T18:13:23Z edgar-rft joined #ccl 2016-07-26T18:57:30Z TwistedFruit is now known as Landru 2016-07-26T19:07:48Z vydd_ joined #ccl 2016-07-26T19:11:45Z vydd quit (Ping timeout: 276 seconds) 2016-07-26T19:13:24Z vydd_ quit (Ping timeout: 260 seconds) 2016-07-26T20:34:59Z vydd joined #ccl 2016-07-26T20:56:51Z pierpa joined #ccl 2016-07-26T21:25:58Z PuercoPop quit (K-Lined) 2016-07-26T21:26:40Z PuercoPop joined #ccl