libera/commonlisp - IRC Chatlog

22:24:20 Guest74 and I thought butlast wasn't such a good name, but now it makes sense.

22:24:43 Guest74 uh, no, i still don't like it.

22:25:11 phoe "the last N elements" versus "everything but the last N elements"

22:25:17 phoe LAST versus BUTLAST

22:26:08 Guest74 as a cobbler it doesn't sound good.

22:28:18 Nilby and there's BUTFIRST as I like to call CDR

22:30:30 Guest74 might as well make a whole bunch.  butnth

22:31:45 Nilby you can rename nthcdr and cd*dr's

22:32:50 Guest74 sure, and replace remove-if-not with but-if

22:32:56 phoe nthcadadr

22:35:01 Guest74 anybody have a jps+ implementation?

22:47:14 jackdaniel nthc(random-number-of-times (random-elt a d))r

22:58:03 Guest74 I just noticed I'm getting a warning on a case slot with :INITFORM nil and :TYPE (or nil something)

22:58:09 Guest74 s/case/class

22:58:21 jackdaniel the type nil is empty

22:58:25 jackdaniel you want (or null something)

22:58:37 Guest74 thanks

22:58:42 jackdaniel sure

23:08:21 rotateq and I wanted to admit my logical failure of yesterday, of course an AND in the matching expression wouldn't make sense ^^ maybe in a quantum computer with quilc :)

23:28:56 temp1234 So, is an fexpr just a macro whose body is implicitly quoted? I've been trying to understand why you'd want them :)

23:29:54 hayley A fexpr is a function which gets its arguments unevaluated.

23:30:22 hayley Theoretically they are more powerful than macros, but I can't think of any uses of fexprs that macros don't do.

23:31:02 temp1234 Hm, alright

23:31:40 White_Flame is a fexpr expcted to actualy execute the code in the runtime environment, instead of returning source code?

23:33:12 hayley Yes.

23:47:45 Nilby I don't think fexprs are more powerful than macros, it's just that given fexprs you can make macros better than the other way around. Most of the time you want macros.

23:51:13 rotateq hm i know mexprs from mathematica and such, what are fexprs again exactly?

23:52:55 Nilby like hayley said, a function that gets un-eval arguments like a macro

23:53:09 mfiano Does the standard provide the building blocks for them, or do they require implementation support?

23:54:31 Nilby You can make a macro that is the fexpr, but then you can't pass it to apply, so you have to make another function which you can apply. So it's a little clumsy, but you can do it in any CL.

23:55:56 mfiano You can apply a macro if it also expands to that second function though, though maybe that is not possible in this specific context.

23:57:55 Nilby It's a tiny bit different because applying the macro function just does the expansion and doesn't evaluate it.

0:09:38 Nilby Let's just forgets fexprs. But this might clear it up or make it more muddy: https://plaster.tymoon.eu/view/2781#2781

0:15:39 jmercouris Bike: exactly what pjb said

0:33:52 _73 I am trying to use the `bst` package to make a binary search tree that can be controlled through a closure. The package however uses special variables to control a trees behavior. I want to be able to set these variable only in my lexical closure so they only control my one instance of a bst. I understand why this code does not work but am wondering if it is possible to do something like this with a closure, or if I need to use a

0:33:52 _73 different package: https://dpaste.com/2C4Z48NRU

0:35:02 Bike have your closure bind the specials on entry

0:35:22 Bike like, you do (let ((bst-copy-function #'identity)) (lambda (...) (let ((*bst-copy-function* bst-copy-function)) ...)))

0:38:29 _73 Perhaps I don't actually know why my code doesn't work. I was under the impression that you could not modify a special variable through a let binding.

0:39:17 Bike yes. let doesn't modify the existing binding, it makes a new one. is that relevant?

0:40:25 _73 how does the bst package know to use this new binding?

0:41:50 mfiano special variables are a stack of bindings under the same name

0:42:40 mfiano You can push new bindings to the stack with LET, and when the dynamic scope ends the value is popped off the stack

0:43:11 Bike well, i'm assuming you'd be using the actual package's special variables here.

0:43:24 Bike since using your own is obviously not going to help

0:45:58 Spawns ** NICK Spawns_Carpeting

0:47:08 _73 I have become confused. I must need to read more about special variables.

1:34:24 karlosz_ ** NICK karlosz

2:00:12 rotateq me too still

2:04:11 rotateq hm maybe someone should give me a hint in the spec error with PROG2

2:09:05 hayley PROG2 should return the primary value of evaluating the second form, but the specification claims it returns the primary value of evaluating the first form.

2:10:27 semz The standard is correct fwiw. It's a typo in the hyperspec.

2:10:29 rotateq ACTION must think a moment just to get it right

2:10:49 rotateq haha yes, i was jsut curious today after the talk

2:13:24 rotateq OK.

2:14:04 rotateq hm now more thinking about the differences fexprs vs macros (or rather sexprs, or maybe you can give me more hints)

2:15:16 hayley fexpr stands for "funny expression" or something like that; it's independent to surface syntax like S-expressions.

2:26:06 rotateq hm okay and mexprs was for meta-expressions. Wolfram still claims that Wolfram Language is the most advanced in existence. but sure they did and do a great job on many things

2:28:03 hayley I'd have no idea how you measure advanced-ness.

2:28:31 hayley But I don't think it has much to do with syntax.

2:29:27 rotateq very impressive, he wrote (much of or everything? hm ok not the existent fortran libs nor based langs) Mathematica 1.0, coming from his SMP, himself

2:30:01 rotateq hayley: right! they claim "we have everything you need in best possible version"

2:32:33 rotateq so to lastly be ontopic with it, maybe when building new things, we can learn from this too and improve it

2:35:34 rotateq and no it hasn't, they have put much know-how into all things over the years

2:39:32 jcowan you want ({drop,take}-{left,right} list n)

2:41:29 rotateq oh do i? :D

2:41:42 rotateq maybe in Haskell such constructs

2:43:57 jcowan Scheme has them too

2:45:04 jcowan https://github.com/scheme-requests-for-implementation/srfi-1/blob/master/srfi-1-reference.scm

2:45:33 jcowan you'd need to modify that if you want portable code

2:45:39 rotateq yes and i read that this newLISP uses define-macro to make them

2:45:43 rotateq yes

2:47:12 jcowan but since almost every CL has tail recursion (if not general tail calling) then it's just a matter of respelling define as defun, etc.

2:47:27 hayley "new", uses dynamic scoping by default

2:49:46 rotateq ah so tail calling is an improvement for tail recursion?

2:51:49 edgar-rft ACTION calls you all tails

2:53:43 rotateq edgar-rft: Sakrileg!

2:54:50 Bike rotateq: i've done some experiments with including fexprs in a compiled language, but it's entirely a pain in the ass, and I don't think it would be an improvement over macros in very many cases.

2:54:59 Bike they are indeed orthogonal to the surface syntax

2:57:42 rotateq oh cool, what didn't you do? :) i see you in so many great projects on github

3:00:04 drakonis rotateq: iirc take what wolfram says with a grain of salt

3:00:34 Bike well, i didn't finish it, for one thing

3:00:54 drakonis i'd be surprised if he did everything by himself

3:01:15 drakonis as he's known for taking credit for things he hasn't built

3:06:01 Nilby I don't how to measure advancedness, but I do know how to measure scientific arrogance, and the unit is the "wolfram" and Mathematica is at about by 37 kiloWolframs by now, which is quite a lot.

3:06:20 hayley I thought arrogance was measured in nano-Dijkstras.

3:06:57 Nilby that's a much smaller scale i think

3:07:12 drakonis nah, wolfram is significantly more arrogant than dijsktra

3:08:08 rotateq drakonis: i like much of his writings and he had much success since from 30 years now

3:08:29 drakonis success built off other people's work though

3:09:35 hayley Well, Wolfram didn't take the offer of using beach's term rewriting compiler, so he can't be that clever.

3:09:43 hayley ("he" being Wolfram)

3:09:49 rotateq and sure, not everything. but the blog article about Schoonschip and nobel prize laureate Martinus Veltman was interesting read. Veltman insisted on writing it in assembly

3:11:04 Nilby unfortunately, his recent physics stuff is closer than most people think, and if people like stylewarning and friends don't realize they're reverse engineering qm into lambda calculus soon, wolfram & co might stumble into it

3:11:07 rotateq oh yes, i think there are so many way more impressive people work for Wolfram Research, just take Oleg Marichev (the author of the five books on sum and integral tables), he did much work so that Mathematica can solve way more integrals

3:11:40 rotateq but okay, when you work at a company they may be allowed to steal your ideas and make billions

3:12:13 rotateq Nilby: oh cool point

3:12:48 rotateq hayley: surely not beach who you meant :) and they corresponded some time?

3:13:17 hayley I forgot the story.

3:13:26 rotateq he could also have written mathematica in lisp, but then just fake symbolity

3:13:36 rotateq no problem, maybe i ask him directly

3:13:38 stylewarning #commonlisp more like #crackpotlisp

3:14:52 Xach Hmm, anyone know more about Harmonix and Lisp?

3:15:08 semz The crackpot is implicit in the Lisp part

3:15:09 Xach I seem to remember at least one CL guy who worked there but I don't know if he did Lisp there.

3:15:43 rotateq ohai stylewarning :) hope the qubits still flip around

3:16:07 stylewarning electrons are definitely bumping

3:16:16 rotateq Xach: hm what is Harmonix?

3:16:21 semz rotateq: If you wait another hour, you'll probably be able to ask beach himself.

3:16:26 drakonis is harmonix as in the people that made rock band?

3:16:51 Lycurgus game maker among many other uses of name

3:17:05 hayley I recall one Wolfram article which went off on Macsyma on Maclisp running on giant timeshared PDP-10s, whereas Mathematica(?) on C ran on a small PC. Never mind the, well, pretty large change in hardware performance.

3:17:09 Lycurgus s/maker/shop/

3:17:17 jcowan rotateq: when foo calls foo in tail position, that's tail recursion. When foo calls bar, that's general ("proper") tail calling.

3:17:24 rotateq stylewarning: I remembered this funny thing from Feynman's auto biography, when his advisor Wheeler called and said "Feynman, I now know why all electrons are the same! They ARE all the same traveling through time and space."

3:17:53 stylewarning The One Electron Universe (TM)

3:18:00 rotateq but should read many parts in the real lectures again the next two years to fresh up my physics too

3:18:24 rotateq stylewarning: and i think this "The Egg" story may have been inspired by it.

3:19:01 hayley ACTION goes searching for the article

3:20:19 rotateq semz: yes he's mostly up around 5AM European middle time

3:21:02 rotateq jcowan: thanks :)

3:22:05 hayley I can't find it now, but I swear someone said some crap like that. Mathematica was released in 1988, which would mean there was literally a 20-year gap in hardware in such a comparison.

3:22:11 rotateq no problem

3:22:36 rotateq but I should sleep, it's now 4:23am here ∘.○

3:23:36 hayley ...apparently the first version was for the Macintosh, so if you had money I guess you'd have a 68030 by then?

3:24:27 hayley (Somewhat amusing is that the 68000 was just used as a "front end processor" which played a part of booting a Lisp machine.)

3:26:25 stylewarning i have a box of like 1-5 FEPs

3:26:27 rotateq :)

3:26:30 stylewarning 15 *

3:26:52 rotateq You also own a lisp machine, don't you?

3:27:04 Guest74 rotateq: I believe Harmonix is the bard.

3:27:10 rotateq hmm

3:27:13 stylewarning i have 2 3650 2 3620 a macivory and a micro explorer

3:27:19 rotateq from Asterix? ^^

3:27:26 stylewarning and a ton of spare parts, keyboards, monitors, etc

3:27:27 rotateq oh wow

3:27:43 rotateq maybe also an extra room for all that :D

3:27:53 stylewarning my guest room is full of it

3:28:07 stylewarning poor guests have to sleep in this literal rats nest of lisp machine ephemera

3:28:12 rotateq haha so guests can sleep on them and it's warm

3:28:47 hayley Oh, I found the comment, and it was from some rando, not anyone from Wolfram. But it's hilarious to me.

3:29:06 hayley "This announcement knocked my socks off. I recall MIT-MC, the multimillion-dollar DEC PDP-10 that ran Macsyma in the late 1970s. It was a scarce resource; labs from all over the world would pay tens of thousands of dollars per year to access the symbolic computation engine. Today, a far more-capable engine is available on a $50 computer."

3:29:18 hayley Sure, you only had...er....45 years of hardware improvements there too, buddy.

3:29:36 rotateq nice

3:29:45 Nilby wolfram was trying to complete with macsyma and first wrote a thing called smp and got mad when the universiy said they owned it, so he quit and re-wrote it as mathematica, running on a tiny mac

3:30:56 Nilby but of course there was cool things on MCL, spawning xcode and CCL

3:31:11 hayley (Well, the PDP-10 was first sold 1968, the Pi 1 was first sold 2012, but with old CPUs Broadcom basically didn't want to throw out...but you get the idea, no?)

3:31:16 Nilby and yes I am a crackpot :)

3:31:20 stylewarning wolfram had a contest to try to decrypt or unlock a copy of SMP

3:31:33 stylewarning this was ca 5-10y ago

3:31:50 Nilby heh, he's still mad about it

3:32:01 rotateq Nilby: i still like the 1988 video with him where he shows everything

3:32:03 hayley Did anyone decrypt it?

3:32:09 stylewarning AFAIK no

3:32:28 stylewarning but you have to engage and sign agreements and whatever

3:32:58 hayley Wait, they had the contest because they didn't have an unencrypted copy?

3:33:07 rotateq and now the S-Combinator challenge

3:33:52 Nilby but sometimes good things come from people being mad universities and quiting to re-write some famous software

3:34:09 drakonis yes...

3:34:10 rotateq hui

3:34:12 drakonis lots of things

3:34:25 ldb It seems wolfram needs GPL

3:34:52 rotateq Nilby: oh yes, very true. escaping too academic and elitary systems

3:35:06 rotateq ldb: but all the money!

3:35:06 hayley stylewarning: I saw someone selling a "retro computer kit" with a 68000 included. Well, that covers the FEP, but what about the rest of the machine?

3:36:52 rotateq hayley: can you please say what FEP means in this context? sry

3:37:14 Nilby Front End Processor

3:37:21 rotateq oki thank you

3:38:30 ldb and the FPU?

3:39:23 rotateq front processor unit?

3:42:02 ldb rotateq: that's floating point unit

3:42:40 rotateq aha!

3:43:58 rotateq that was what Veltman also suggests for using assembly in computer algebra, that floating point ops are fasta. but physics also lives heavily from symbolic calculation

3:44:28 rotateq okay, it was another time (or not?)

3:45:36 rotateq stylewarning: hm iirc i should go on and install gfortran on my new machine, so i can finally compile/load the qvm and try playing with it :)

3:48:43 Guest74 rotateq: I have to admit I was incorrect.

3:49:57 stylewarning rotateq: I don't think gfortran is needed with latest master

3:50:32 stylewarning (I might be wrong; I have it installed all the time)

3:51:11 stylewarning But we refactored MAGICL to not need that stuff necessarily https://github.com/quil-lang/magicl#requirements

3:53:47 rotateq okay I don't remember what it let fail the last time i tried, but I want understand some day how this and quilc really works and does its impressive magic, so high respect for you people at rigetti

3:56:25 rotateq hm i wonder if some maybe verbose (from APL view point) things in magicl can also replaced and made shorter with april, even when april becomes in some more things superior over the dyalog overoptimized implementation

4:00:51 beach Good morning everyone!

4:01:30 Nilby the lisp around the old fortran is both funny and sad. there must be a more palatable way

4:04:31 rotateq Hello beach, don't worry, that bit offtopic may now be over. But of course interesting too.

4:05:54 ldb weird, cl-cpus::*number-of-processors* is 0 on OS X

4:08:38 ldb ah, they hardcoded sysctl name constants, and assumed OS X has the same constant as BSD

4:09:06 rotateq beach: Did you correspond at some point with Stephen Wolfram in the past?

4:10:24 rotateq ldb: maybe OS X is so well designed as they often claim, it can also run multithreaded on 0 cores

4:10:24 beach No, hayley forgot the details. It was not *my* term rewrite system, it was that of my thesis advisor, and he was the one doing the communication before I got involved. But I wrote the compiler for that term rewriting system, something nobody had done before.

4:10:46 rotateq Okay, sure also some days ago. :)

4:16:48 rotateq I maybe wasn't even born nor thought of at that time.

4:18:38 beach I started my dissertation in 1983.

4:19:55 rotateq okay then 11 years before i was crafted :)

4:20:33 rotateq you was deeply into lisp at that time, right?

4:20:39 rotateq *were

4:21:55 beach Yes. I wrote the compiler for the rewrite system in Franz Lisp (not to be confused with Allegro Common Lisp).

4:22:07 beach It generated VAX assembler.

4:22:16 rotateq I wouldn't have confused it. :)

4:22:27 beach Good. :)

4:22:28 rotateq hui okay

4:23:30 rotateq but I learned this year that FranzInc is named after the hungary componist, I like some of his music

4:23:49 hayley beach: Thanks for the clarification.

4:24:03 beach Yes, Franz Lisp is a pun on Franz Liszt.

4:24:07 rotateq Franz Liszt

4:24:21 rotateq ah yes you were faster

4:33:42 rotateq so CL was at that time heavily influenced by Maclisp, Zetalisp and others, right?

4:34:03 rotateq or Interlisp and your mentioned Franz Lisp

4:36:23 beach Right. I don't think Franz Lisp had any influence on Common Lisp. Franz Lisp was mostly like Maclisp. It was specifically written for Unix as I recall.

4:38:19 rotateq okay I have no idea, but am interested in those history details. one book of Herbert Stoyan I found some weeks ago from around 1980 covers over 100 pages just LISP history :)

4:38:40 beach Yeah, it's fascinating.

4:51:37 rotateq Can I pass a lambda-expression directly inside a SATISFIES with DEFTYPE?

4:51:52 ldb yes

4:52:29 rotateq can you give me a simple example? and then how to correctly handle (optional) type parameters

4:53:14 beach clhs satisfies

4:53:14 specbot http://www.lispworks.com/reference/HyperSpec/Body/t_satisf.htm

4:53:33 beach It has to be a symbol.

4:54:05 beach "lambda expressions are not allowed"

4:54:22 rotateq oh hm okay

4:54:28 ldb oh, i thought that's argument list

4:58:14 rotateq and which additional type specifiers I remember are defined by default are: satisfies, mod, eql, complement, and, or, not; which did i miss?

4:58:24 clone_of_saturn ** NICK saturn2

4:58:27 rotateq member

5:01:47 beach clhs 4.2.3

5:01:47 specbot Type Specifiers: http://www.lispworks.com/reference/HyperSpec/Body/04_bc.htm

5:02:04 beach COMPLEMENT is not one.

5:02:11 rotateq stylewarning: okay it was about some libffi thingy so file it can't find on my machine, not gfortran

5:02:47 rotateq beach: oh! so good i asked, or should have asked "which of those did i remember correctly?"

5:03:49 rotateq And what is needed for writing efficienty TYPEPs?

5:05:36 beach An implementation that recognizes a call to TYPEP with a quoted type specifier and handles it efficiently.

5:05:50 rotateq mysterious :)

5:06:26 rotateq better maybe: how do you approach it in SICL?

5:06:51 beach You would want your implementation to turn (TYPEP ... 'CONS) into (CONSP ...) for instance.

5:07:54 rotateq yes, so with a compiler-macro?

5:08:08 beach Possibly. Or have the compiler handle it specially.

5:09:28 rotateq I see. Like when SBCL for example sees an addition with two arguments and calls #'sb-impl::two-arg-+ on those.

5:10:10 rotateq Or as is in heisigs fast-generic-functions the example with first binary-add and the reducing more arguments with that.

5:11:29 saturn2 you can consider the second argument to TYPEP as a little programming language which you compile into lisp

5:12:57 rotateq hmm ok

5:50:49 clone_of_saturn ** NICK saturn2

6:23:10 ldb Well I pushed a PR to cl-cpus to fix problem on OS X

6:24:03 ldb but I think april also need a fix, it might be to confident to assume everyone gets 2+ CPU cores.

6:26:42 ldb ah, my version seems outdated, the master branch on github uses shell command now.

10:18:29 Lord_of_Life_ ** NICK Lord_of_Life