freenode/lisp - IRC Chatlog

3:56:28 dyson ** NICK Trioxin

4:29:44 aeth hmm, it looks like I run into the sb-ext:*inline-expansion-limit* if I declaim inline too much (the error actually asks me if I'm inlining something recursive)

4:32:42 aeth It's because I have little inlined functions for (aref foo 0) (aref foo 1) etc and then a function that uses a lot of those that is in turn inlined, that is in turn used multiple times in the functions that give the warning. I comment out the inlines for those tiny functions and it's gone.

4:33:05 Bike huh. would have expected it to go by depth.

4:33:35 aeth https://gitlab.com/zombie-raptor/zombie-raptor/blob/2f7c9d2bf19c7b275ff93a6e4e7c22fbdfe48848/math/quaternion.lisp#L69-75

4:33:38 aeth It's because of that

4:33:52 aeth That itself is called twice in quaternion-rotation-of-vector

4:34:11 aeth which is called 4-5 times in model.lisp, where the errors happen (but only one of the random calls per function)

4:35:00 Bike 32*2*4, there you go then

4:35:08 Bike you could just bump it up obvs

4:35:22 aeth Or I could use a different thing for the arefs? Like a macro?

4:35:36 aeth But generally I thought it was better to use inlined functions for trivial functions

4:35:44 Bike it is, yes

4:36:19 Bike i'd just bump the limit, this seems legitimate to me

4:40:03 aeth Where do I bump it in the program?

4:40:03 |3b| * probably wouldn't inline functions big enough to hit the inline limit

4:41:28 Bike uh... not sure

4:42:03 |3b| asdf tricks to bind it around compiling/loading your files?

4:43:12 aeth |3b|: unfortunately, quaternion* is probably something that needs to be inlined for performance reasons (and is probably the largest one that has to be)

4:43:35 Bike you could also bind all eight values so you don't do a call more than once

4:43:38 aeth Inlining my mostly-trivial quaternion functions got rid of pretty much all of my game loop allocations

4:43:46 Bike which would get you like 64 inlines overall

4:43:50 aeth Bike: yes, that's what I did in a previous version, I had a let statement

4:43:53 |3b| function call is relatively small part of a function that does a lot, have you profiled?

4:44:08 |3b| if it is for allocation that is harder to fix :/

4:44:15 Bike i dunno if this function does a lot, it's just arithmetic

4:44:22 aeth |3b|: I'm guessing when I inline the quaternion allocations, SBCL is smart enough to automatically (declare (dynamic-extent some-quaternion)) which does help a lot

4:44:49 |3b| Bike: well, i guess it if has lots of layers of inlining

4:44:52 Bike yeah i don't think sbcl knows how to dynamic extent anything that isn't a direct make-array or make-struct or whatever

4:45:12 |3b| * wouldn't expect it to unless you tell it to

4:45:14 Bike though that wouldn't matter to this

4:45:31 Bike maybe inlining here saves boxes?

4:45:50 |3b| * wouldn't expect that either with single floats on x8664

4:45:59 Bike oh is it singles

4:46:14 Bike tis. hum.

4:47:34 aeth Bike: I have two versions of most functions, some allocate into an existing vector/quaternion (usually -into!) and some allocate a fresh vector/quaternion

4:54:36 aeth I did rewrite some to take advantage of inlining, though, (and reduce repitition) so it looks like there is more allocations than there were

4:56:16 aeth I think what really got rid of most of the allocations, though, was inlining the quaternion function that makes the quaternions themselves

4:56:42 aeth i.e. #'quaternion (if I wasn't clear)

4:57:31 Bike yeah cos like i said http://www.sbcl.org/manual/#Dynamic_002dextent-allocation "SBCL implements stack allocation for"

4:59:32 aeth Right, direct make-array as you said.

5:01:22 aeth The rest of the allocations that went away were probably due to something like it not being aware of a type somewhere or something.

5:02:20 Trioxin does anyone have a long-running program that's constantly searching the entire programming space in some fashion? I remember ray kurtzwiel once say "I could write a superintelligence in 50 lines of Lisp." Of course it would take more computational power than we have for that to be worth anything but it would still be a cool thing to have running.

5:02:42 Bike too big, too long.

5:04:04 aeth If I could write a superintelligence, I could write a superintelligence in 1 line of Lisp. Mainly because if it can fit in one top-level form it can be put all on one line (but shouldn't) :-p

5:04:50 aeth Someone should start an International Obfuscated Common Lisp Superintelligence Contest one day.

5:05:20 fiddlerwoaroof Yeah, a superintelligence is more defined by the speed it gets an answer than whether or not it will eventually get an answer

5:05:28 Trioxin obfuscation/minifying wasn't the point of it

5:05:32 Bike and an AI isn't exactly a mu recursive function so a normal thing like interleaving computation would be meh

5:07:07 aeth Trioxin: I've heard exactly the opposite from some people. That we have the computational power already, but we don't have the algorithms.

5:07:42 aeth Although I guess it's not quite the opposite.

5:07:51 Bike the blind search would take more "computational power" than we have, insofar as that means anything

5:07:56 Bike i assume that is what was meant

5:07:57 Trioxin aeth, that's a separate thing. what he meant was a program that's just randomly trying different arrangements of code

5:08:19 aeth Infinite monkeys writing Shakespeare

5:08:30 Trioxin right lol

5:08:37 PuercoPop jasom: The following snippet from the reference of Parenscript does not match the documentation. Would this be considered a bug? (ps:ps (new (Person age shoe-size)))

5:08:42 Bike worse, infinite monkeys trying to build the playwright from scratch

5:08:43 Trioxin but you know.. maybe it would get lucky

5:08:57 aeth In case anyone reading isn't familiar: https://en.wikipedia.org/wiki/Infinite_monkey_theorem

5:09:00 Bike and they're finite and you have to pay their food bills

5:09:08 Trioxin sooner rather than later

5:09:29 Trioxin like winning the cosmic lottery

5:09:44 aeth The problem with infinite monkeys with infinite typewriters is... infinite is huge. No, larger than that. No, even bigger.

5:09:50 Bike finding the sun out of all the stars that exist would be easier

5:11:40 Trioxin unless you could define some very clever fitnesses

5:11:49 Trioxin maybe

5:11:54 aeth Trioxin: well you could just have humans read the programs, and maybe it'd inspire someone if it's close

5:12:43 Trioxin some autistic programmer

5:13:59 aeth I thought of something a long time ago about just generating an infinite amount of images. You could simplify by e.g. only having it be black or white and you'd probably still get some meaningful images generated.

5:14:30 aeth That's an easier thing to do, and humans can visually verify things much faster.

5:15:05 Bike you should really try that to get some perspective on "probably"

5:15:05 Trioxin we could code something to pick out the interesting ones these days

5:15:35 Trioxin black and white 256x256

5:15:55 aeth Bike: almost everything will be static noise (although humans will still see patterns in those), but you'll also find everything that can be expressed at that image size, as long as it's small enough that you can actually generate them all

5:16:03 aeth so you'll see e.g. mickey mouse

5:16:36 Trioxin you'll see a miniature likeness to the REAL god

5:17:00 Bike i mean, there are, let's check the old calculator here, 10^19728 ish 256×256 black and white images

5:17:49 Trioxin alright 50x50

5:17:50 Bike so you can save a lot of time by just looking for toast with christ's face in it.

5:18:13 aeth Bike: that's why I didn't specify the image size, only the 1 and 0 color restriction

5:18:23 Bike 50×50 is still 10^752.

5:18:38 Trioxin meh

5:18:51 Bike brute force search is dumb. that's why "brute" is in it.

5:18:52 Trioxin what's the size they use for imagenet?

5:19:21 aeth Trioxin: right, the advantage is you'd have to go so small that deep learning could handle the images! :p

5:19:44 aeth So maybe you brute force generate the images, feed them into deep learning to eliminate noise, and then look at what you brute force generated?

5:19:50 aeth not sure what the point would be, though

5:20:10 Trioxin you'd have a ton of logo ideas

5:20:44 aeth you could also find a way to filter things that will mostly look like static ahead of time, somehow

5:21:29 Trioxin hmm... that might be more expensive than it's worth

5:21:29 Trioxin the filtering

5:21:32 aeth you'd probably be looking mostly for silhouettes

5:21:41 Trioxin well, no.. convnet more expensive

5:22:38 Trioxin everyone's face would be in it

5:22:56 aeth Bike: one advantage to the dumbest approach, though, is that it's embarassingly parallel

5:23:44 Trioxin maybe i'll code it for fun. will have to use the fastest thing I know (Nim)

5:23:45 aeth you don't need to just stick it in (loop for i from 0 below (* 50 50) do ...)

5:24:14 contrapunctus Does nobody like minispec.org?

5:24:16 Bike great, now you can go through the cosmic search space a thousand times as fast

5:24:37 contrapunctus (seemed like a nice project, but also seems dead...)

5:25:03 Trioxin i could get my GPU on the job

5:25:19 aeth Bike: Possibly up to a million so you can just go 10^752 / 10^6 = 10^(752-6) = 10^746

5:25:26 aeth hah!

5:25:41 aeth (that's if you wanted to use a seti@home style approach)

5:26:04 Trioxin well I'm sure I'd have it computed in my lifetime

5:26:18 Bike no.

5:26:31 Trioxin assuming performance gains

5:26:36 aeth and if we can't solve the simple image problem, how can we create the Lisp program?

5:27:36 Bike if you did a quintillion images every attosecond it would still take 10^716 seconds to get the entire search space. it is a dumb problem.

5:28:11 Trioxin my quantum computer will do it

5:29:18 Trioxin how many qbits to solve it in a year?

5:30:14 Bike quantum computers aren't magic.

5:31:05 Trioxin they are until I decide to understand them better

5:31:16 Bike gross.

5:31:27 aeth what if I use this? (declare (optimize (speed 3) (safety 0))) /s

5:31:37 aeth That probably shaves off trillions of years, actually

5:34:19 Trioxin quantum computers are very uninsteresting when you have no hope of accessing one

5:34:51 Bike what a terrible attitude. lots of scientists can't touch anything they work with, you know.

5:34:53 Trioxin of course you can make that statement sound really stupid with a good analogy

5:36:13 Trioxin lets see where's that good thing

5:36:20 Trioxin google*

5:36:50 Trioxin http://www.quantumplayground.net/

5:44:42 aeth I inefficiently solved the image problem for a 2x2, btw: (let ((v (make-array 4 :adjustable t :fill-pointer 0))) (dotimes (i (* 2 2) v) (vector-push-extend (read-from-string (format nil "#*~2,'0B" i)) v)))

5:48:03 Trioxin nice

5:48:16 Trioxin getting matt cutts on the phone

5:52:39 aeth this is how it scales: (let ((v (make-array 50 :adjustable t :fill-pointer 0))) (dotimes (i (expt 2 (* 4 4)) v) (vector-push-extend (read-from-string (format nil "#*~16,'0B" i)) v)))

5:53:28 aeth so I'm getting (expt 2 (* 50 50)) for the 50x50 version... and (log (expt 2 (* 50 50)) 10) => 752.575

5:54:56 aeth If you want to run out of memory: (let ((v (make-array 50 :adjustable t :fill-pointer 0))) (dotimes (i (expt 2 (* 50 50)) v) (vector-push-extend (read-from-string (format nil "#*~2500,'0B" i)) v)))

5:55:19 aeth There are far more efficient ways of doing it, but if Bike's right you won't avoid the (expt 2 (* 50 50)) so it won't matter

5:56:12 Bike and i'm never wrong

5:59:03 aeth If you wanted to be smart with memory, you'd probably make it all one big simple-bit-vector and just make each image a subseq

6:00:33 aeth That will limit what you can handle in RAM at once to array-dimension-limit

6:00:55 aeth in SBCL, 4611686018427387901

6:03:40 aeth That's 3 bytes less than (expt 2 62) or (expt (expt 2 10) 6.2) so about 4 exbibytes of RAM.

6:10:20 aeth That's not enough to fit the whole thing, which would afaik require (* 2500 (expt 2 2500)) which afaik means it'd need a unit at 1024 to the (log (* 2500 (expt 2 2500)) 1024) => 251.12877

6:11:06 aeth the name for the unit of (expt 1024 251) bytes hasn't been invented yet, afaik.

6:13:37 Bike doesn't come up much

6:15:04 aeth If I modelled it right, though, 6x6 does fit in contemporary RAM with just (* 6 6 (expt 2 (* 6 6))) bytes required so (/ (* 6 6 (expt 2 (* 6 6))) (expt 1024 4d0)) => 2.25d0 TB

6:15:34 aeth and (/ (* 5 5 (expt 2 (* 5 5))) (expt 1024 3d0)) => 0.78125d0 GB

6:18:32 aeth so I guess stick to 5x5, which is much closer to what my intuition thought (it doesn't look like I actually named any numbers, I too was afraid of Bike to do that)

6:19:53 smokeink is it possible to "hook" at higher level and make a restart that's triggered inside a form execute it's original code + some more ?

6:20:13 drmeister Bike: Have you had any interaction with the METER code in Cleavir?

6:20:17 Bike some more that's provided by the handler?

6:20:19 Bike drmeister: only a little

6:20:27 drmeister Is it always on?

6:20:39 Bike iiiii think yes

6:20:41 smokeink some more that i provide at higher level

6:20:43 drmeister I'm trying to track down a 30-50% slowdown in building clasp

6:21:07 Bike smokeink: you could have the handler pass the restart a function

6:21:09 drmeister It's like everything takes 1.5x longer and uses 1.5x more memory. It's very frustrating.

6:21:18 Bike with arguments, if you need stuff in the lexical enviroonment

6:21:28 Bike drmeister: i don't think any meters have been touched recently

6:22:17 smokeink the thing is this restart is inside a library , it doesn't accept a function as it's parameter

6:23:14 Bike then basically no. the definition of the restart controls what happens.

6:23:17 Bike what are you trying to do?

6:23:39 smokeink is it possible at higher level to add an extra restart that if chosen by the user then it will execute a restart defined in the lib + some extra code given by me at higher level ?

6:23:52 smokeink * 1 sec i'll make a short paste

6:24:00 Bike add a restart -> no.

6:28:43 AntiSpamMeta2 ** NICK AntiSpamMeta

6:32:12 smokeink http://pastecode.ru/18b26a/

6:32:27 whiteline_ ** NICK whiteline

6:33:15 gendl_ ** NICK gendl

6:33:23 velvetcore_ ** NICK velvetcore

6:33:26 Neet__ ** NICK Neet_

6:39:52 smokeink hmm, im gonna try to wrap it in a handler-bind and add a restart-case inside handler-bind's lambda function

6:42:10 Bike can you not just call checkl-store from inside the handler?

6:43:26 smokeink i want to call checkl-store only when use-new-value restart is chosen, not every time the error is signaled . Because sometimes the error might be due to a real bug in the functions so i don't want test results to be updated to some wrong ones

6:44:11 Bike uh... so do it only in the handler that does use that restart?

6:45:18 smokeink can you modify my paste to show me your point? add some pseudocode

6:46:17 smokeink i'm not sure what/which the 'handler that uses that restart' is

6:46:43 Bike wait, are you j ust choosing restarts from the debugger

6:47:02 smokeink yes

6:47:36 Bike in slime you can just hit 'e' to evaluate something. you can just type in the checkl-store form.

6:48:18 smokeink yes that 's an option

6:48:29 smokeink but to have it automated is even nicer

6:48:58 Bike so you want a level of optimization where you still select the restart manually, and it does that for you?

6:49:14 Bike s/optimization/automation/

6:49:16 smokeink yes

6:50:02 smokeink the only manual action is to select use-new-value restart, and i want it automatically to save the results to file

6:52:01 smokeink use-new-value just updates the values in memory, it does not save the values to the file

6:53:21 Bike (handler-bind ((checkl::result-error (lambda (c) (restart-case (error c) (use-new-value-and-save (new-value) (checkl-store #p"/test/values.txt") (invoke-restart 'use-new-value new-value)))))) (check-formal or whatever))

6:53:32 Bike that's twisty enough that i dunno if it will work without testing

6:54:14 smokeink :) thanks, i think it'll work, that's exactly what i was about to try

6:55:55 smokeink didn't know about the (restart-case (error c)) trick though

7:07:05 smokeink use-new-value doesn't have a param , checkL uses a global var. Now the only problem is that (invoke-restart 'check::use-new-value) has to be evaled before (checkl-store) so that the value in the global var is updated before it's dumped out to the file.

7:07:57 smokeink unfortunately it seems that forms after (invoke-restart) are not executed any more

7:14:13 myrkraverk_ ** NICK myrkraverk

7:16:15 Bike uh... yeah, invoke-restart transfers control

7:16:51 Bike back to the code you don't control, particularly

7:17:00 Bike so having to do it afterward takes us back to "no"

7:18:30 smokeink damn, this one transfers control as well (checkl-store

7:18:34 smokeink (funcall (slot-value (find-restart 'checkl::use-new-value c) 'function)) <- this one

7:18:51 Bike don't do that

7:19:00 Bike that's the same as invoke-restart anyway

7:19:17 Bike there is no way to invoke the restart without transferring control, because that doesn't make sense

7:20:36 myrkraverk_ ** NICK myrkraverk

7:21:46 beach Good morning everyone!

7:22:32 smokeink good morning

7:23:37 smokeink bike: yeah but it'd be cool to have 'restart' (the transfer of control) and 'restart-body' (the code before transferring control) separate

7:24:21 Bike restarts are usually intended to run in the context they're written in

7:24:37 Bike you can have a restart that RETURNs something, for instance

7:25:30 Bike nevermind, that's not a good example

7:27:41 smokeink so can we say that checkL has a design issue , since we can't hook into that restart and automatize stuff ?

7:28:12 Bike i wouldn't say so... you're asking for quite a bit of control over what checkl does

7:52:28 smokeink ok i understand

8:20:13 shaftoe is there a way to see available slot-values or accessors on a class instance object?

8:20:26 shaftoe ie, like (hash-table-keys/values ..)

8:22:34 flip214 shaftoe: just inspect it via swank

8:22:46 flip214 clhs inspect

8:22:46 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_inspec.htm

8:23:17 flip214 clhs describe

8:23:18 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_descri.htm

8:23:21 flip214 shaftoe: ^^

8:23:57 shaftoe cheers

8:28:55 contrapunctus Just what am I doing wrong? (cl-ppcre:regex-replace-all "\\p{IsGraph}" "abc" "X") => "abc" (rather than "XXX"; cl-unicode is loaded)

8:29:00 smokeink swank can inspect functions and display the assembly code, can the asm code be examined without swank ?

8:36:44 edgar-rft clhs disassemble

8:36:44 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_disass.htm

8:37:27 smokeink nice

8:37:32 edgar-rft :-)

8:37:48 shaftoe it seems like most things you could want exist

8:38:50 edgar-rft not all, but many

8:42:26 contrapunctus anyone?

8:53:05 flip214 contrapunctus: you did load :cl-ppcre-unicode? what else did you load?

8:54:10 flip214 contrapunctus: does it work with any other property names?

8:55:09 jdz contrapunctus: also, are you sure "IsGraph" is a category cl-unicode recognises?

8:55:09 contrapunctus flip214: oh, wait. (ql:quickload :cl-ppcre-unicode) -> http://ix.io/1STq

8:55:22 flip214 (cl-ppcre:regex-replace-all "\\p{ASCII_Hex_Digit}" "abcdefghijkl" "X")

8:55:26 flip214 does work for me...

8:56:14 flip214 contrapunctus: is the installed cl-ppcre from a system package? "apt-get remove cl-ppcre" and then "(ql:quickload :cl-ppcre")

8:57:47 jdz contrapunctus: I don't see IsGraph here: http://www.unicode.org/reports/tr18/#General_Category_Property

9:00:39 jdz And yes, when cl-ppcre uses cl-unicode, it complains about unknown category IsGraph.

9:04:31 contrapunctus ah, thank yyou, flip214 and jdz

9:05:41 contrapunctus flip214: your example returns "XXXXXXghijkl" here, for some reason

9:06:27 jdz Isn't that the expected result?

9:07:22 contrapunctus I'd like to replace any and every graphical character (non-whitespace)

9:09:13 jdz I don't think it's that simple.

9:10:12 contrapunctus jdz: How does one go about it, then :\

9:10:48 phoe_ clhs whitespacep

9:10:48 specbot Couldn't find anything for whitespacep.

9:11:30 phoe_ clhs graphic-char-p

9:11:30 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_graphi.htm

9:11:34 jdz contrapunctus: it depends on what you are _really_ after. For instance, check http://www.unicode.org/reports/tr18/#Compatibility_Properties, the "space" row.

9:11:57 phoe_ contrapunctus: (loop for char across string unless (graphic-char-p char) do (setf char #\X))

9:12:11 phoe_ clearly not regex enough but oughtta work

9:12:30 beach clhs substitute-if

9:12:30 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_sbs_s.htm

9:12:38 contrapunctus jdz: I'd like to replace "\([:graph:]\) with ``\1, \([:graph:]\)" with \1''

9:12:44 hhdave_ ** NICK hhdave

9:12:56 jdz contrapunctus: define ":graph:".

9:13:08 contrapunctus jdz: non-whitespace

9:13:18 jdz contrapunctus: define "whitespace".

9:13:20 contrapunctus I could just hack it and [A-Za-z0-9] or something

9:13:38 contrapunctus jdz: spaces and tabs...oh.

9:14:04 jdz contrapunctus: what about " "?

9:14:33 jdz What about "​"?

9:14:50 loke And what about " "?

9:14:51 contrapunctus the former is graphical...the latter is...an empty string?

9:14:55 jdz What about vertical tabulation?

9:15:22 loke Found a nice list: https://www.cs.tut.fi/~jkorpela/chars/spaces.html

9:15:25 contrapunctus oh, that's..."non-breaking space"?

9:15:25 jdz contrapunctus: no, it contains a glyph.

9:15:47 jdz There's a zero-width-space in "​"?

9:15:54 jdz That's not a question.

9:16:07 loke There is a HAIR SPACE in my example.

9:16:30 loke There is also IDEAOGRAPH SPACE: "　"

9:16:35 loke And several more.

9:17:08 loke jdz: Your NO-BREAK SPACE probably shouldn't be seen as a separator...

9:17:13 jdz «The characters in \p{gc=Format} share some, but not all aspects of control characters. Many format characters are required in the representation of plain text.»

9:17:55 jdz contrapunctus: anyway, welcome to Unicode, enjoy your stay.

9:18:42 contrapunctus lol

9:18:50 jdz contrapunctus: http://www.unicode.org/reports/tr18/#Compatibility_Properties also has a "graph" row -- maybe that's what you want?

9:19:47 contrapunctus a different (less robust?) way of doing the same thing, I guess, is - (cl-ppcre:regex-replace-all "\(^\| \| \)\"" "\"abc\"" "\\1``")

9:20:57 jdz contrapunctus: if you just want to substitute particular characters then you don't even need regular expressions (and you've been given example code already).

9:21:39 loke contrapunctus: You probably want to look at the function SB-UNICODE:WHITESPACE-P

9:21:41 contrapunctus jdz: in "abc", I need to change the first " to ``, then second to ''

9:21:54 loke contrapunctus: that's horrific.

9:22:01 iago clisp noob here, but if you want to match everything but withespaces wouldn't be easier to negate a char class ? [^\s\n\t\r]

9:22:05 contrapunctus loke: which one? :(

9:22:05 loke contrapunctus: You want to be using “ and ” (in English)

9:23:09 contrapunctus loke: I'm writing a script to convert Wikisource books to LaTeX; some of them use "", some “”

9:23:27 jdz loke: there are other languages than English, you insensitive clod!

9:23:31 loke contrapunctus: You only have to deal with ".

9:23:42 loke contrapunctus: You can put “ and ” directory into LaTeX source.

9:23:47 loke I mean _directly_

9:23:52 jdz In Latvian, for instance, it's „ and ”.

9:23:59 beach iago: CLISP is an implementation of Common Lisp. So we don't use CLISP as an abbreviation for Common Lisp. Use "CL" if you must abbreviate.

9:24:04 loke In Swedish, you do it ”like this”

9:24:45 iago beach: ok sorry

9:24:47 flip214 I say, «this is another quoting»!

9:24:49 jdz loke: two "right quotation marks"?

9:24:58 loke I believe in German, it's « Like this ». Not the HAIR SPACE surrounding the text.

9:25:08 loke jdz: Yes.

9:25:21 jdz loke: that's convenient.

9:25:23 contrapunctus I thought it was „this” in German

9:25:53 loke jdz: Look at the headline on this Swedish news site for an exmaple: this

9:25:57 loke http://www.dn.se/

9:26:02 jdz Actually, both „” and «» are "valid" in Latvian, and those might be because of the German influence.

9:26:28 jdz loke: I would not use web pages as a measure of correctness.

9:27:03 loke jdz: That's the largest newspaper in Sweden, and has traditionally been the bearer of language correctness for the last 100 years.

9:27:20 loke jdz: Trust me, they're doing the quotes right :-)

9:27:51 jdz loke: I see, and that's nice.

9:29:09 loke it's probably because I grew up with it, but I feel the ”right quote” style is much nicer than the “upside-down” style. But... I use it when writing in English, of course.

9:30:08 loke What is not entirely clear to me is how to use ‘single’ vs. “double” quotes in English. In Swedish it's much easier: You ”use then when you have a ’recursive’ quote”.

9:30:17 flip214 speaking as someone who's doing erratas from time to time, having matched quote pairs is just another thing that goes wrong. Even if it's nicer to read ;)

9:30:32 loke flip214: You make a fair point.

9:31:07 loke For that, the « angle style » should win :-)

9:31:11 flip214 these should just fall out correctly by using some sane text processing system... but most people don't.

9:31:29 flip214 loke: yeah, exactly.

9:31:55 flip214 and for quoted quotations there's ‹› too

9:31:55 loke flip214: True. I've mapped all of these quotes to my keyboard so I can use them correctly. Easier than crossing your fingers and hoping your mail client does the right thigng.

9:32:17 loke flip214: Oh nice! I don't think I have it on my keyboard.

9:32:28 jdz loke: what, you don't use Emacs to write your mails?

9:32:33 loke jdz: I do.

9:33:00 loke jdz: And Emacs is one of those applications that doesn't even try to be clever in that respect. The other way I type emails is in Gmail.

9:33:10 contrapunctus smartparens!

9:33:11 TMA except that the »angle style« is reversed in certain places also „“ is used

9:33:20 loke TMA: Wait what? Where?

9:33:21 loke :-)

9:33:31 jdz But yeah, I also have them mapped in my keyboard layout (which I've had to create manually anyway, since there is no Dvorak/Latvian out of the box).

9:33:55 loke jdz: Does latvian use äö?

9:33:58 TMA loke: Czech (there are single quotes in use too, but less so)

9:34:02 loke or ü?

9:34:04 jdz loke: no, just āē :)

9:34:21 loke Nice.

9:34:31 loke jdz: I had no idea you were latvian.

9:34:32 jdz loke: we have 3 accents: macron (ā), caron (č), and cedilla (ņ).

9:34:50 loke jdz: Ah, they are accents? Not letters in their own right?

9:35:15 jdz loke: well, accented characters, and they are part of the alphabet as separate letters, yes.

9:35:40 loke jdz: Ah.

9:35:59 jdz Anyway, I think we've now gone waaaay off-topic now.

9:36:23 loke jdz: Does it matter

9:36:24 loke ?

9:36:44 jdz Not until somebody complains.

9:36:55 phoe_ (čar '("accents" . "matter")) ;=> "aččents"

9:38:12 TMA we have three diacritics too: acute (á), caron (č), and ring (ů); there is also sometimes an umlaut in german loanwords too

9:39:56 phoe_ ąćęłńóśźż

9:40:05 phoe_ ĄĆĘŁŃÓŚŹŻ

9:40:08 phoe_ that's for Polish

9:40:14 flip214 phoe_: be careful with that czar.

9:40:22 flip214 ;P

9:40:22 phoe_ flip214: czar?

9:40:46 flip214 well, how is ĉåŕ spoken?

9:41:04 TMA phoe_: 1. it's čář 2. ;=> "áččěňťš" ; čďř is "mäťťéř"

9:41:39 phoe_ TMA: ah, thanks

9:43:05 TMA áčďéěíňóřšťúůýž (and äöü in german loanwords)

9:44:34 flip214 next thing here will be colors and ascii-art

9:46:31 TMA Slovak has áäďéíĺľňôŕšťúýž

9:48:32 TMA flip214: ĉ is like the English ch (esperanto), å has its swedish sound and ŕ is long vocalic r, therefore the ĉåŕ is disyllabic: ĉå-ŕ :)

9:49:08 flip214 TMA: how much difference to "czar"?

9:49:44 TMA flip214: altogether it would be probably mispronounced as čór which is Roma word for thief

9:51:20 TMA flip214: quite differently

9:55:32 TMA flip214: fitst, there is no palatalization of the initial consonant in czar (palatalization is the difference between s and sh (try the unpalatalized sit versus the palatalized variant; observe your tongue tip position)

9:56:06 flip214 TMA: thank you very much. but it was meant as a joke, so this is too many details. sorry.

9:56:19 TMA flip214: yw :)

10:15:42 phoe_ aw geez

10:15:45 phoe_ time for another mammoth

10:16:05 phoe_ * starts hunting the page named Macro PROG, PROG*

10:16:14 phoe_ (I've never used these macros)

10:17:03 phoe_ PROG is basically BLOCK NIL + LET + TAGBODY, I see

10:17:49 phoe_ and PROG* is basically BLOCK NIL + LET* + TAGBODY

10:39:22 phoe_ haha, the example

10:39:30 phoe_ KING-OF-CONFUSION versus PRINCE-OF-CLARITY

10:40:00 phoe_ the everlasting battle that no one I know watches because I don't see PROG or PROG* used in the code that I read.

10:42:02 phoe_ haha! I'm editing the page for PROG2. I can finally fix the error on it.

10:42:13 contrapunctus perspicuously

10:52:05 phoe_ I love it

10:52:25 phoe_ PROGN takes forms as input, are what are forms? They're an implicit progn.

10:52:42 phoe_ So every PROGN form has two parts: an explicit progn and an implicit progn.

10:52:52 lieven phoe_: are you rewriting the hyperspec?

10:53:01 phoe_ You can't make an explicit progn without an implicit progn.

10:53:02 phoe_ Haha.

10:53:05 phoe_ lieven: sort of.

10:53:12 phoe_ http://phoe.tymoon.eu/clus/doku.php?id=clus:todo

10:55:16 phoe_ it's a good approximation

10:55:30 phoe_ although I'm not using the HyperSpec because of its licence.

10:55:53 lieven and what are your goals? clearing it up or working towards a new standard?

10:57:22 phoe_ lieven: clearing it up, formatting, hyperlinking, and exposing to the Lisp community in a clean, expandable and editable format.

10:58:15 lieven one of the major things I always disliked about it is that some fairly important stuff is hidden in the glossary

10:58:33 phoe_ lieven: https://github.com/phoe/clus-data

10:58:40 phoe_ describe your issue and file it there.

10:59:06 phoe_ once I finish roughlu converting the specification, we (as a community) will be able to address the problems with its current form.

10:59:38 phoe_ and I mean it. the more issues we have there, the more material we will have for building better documentation for CL.

11:01:18 lieven I've forgotten the details but will do if I remember

11:05:53 phoe_ lieven: do it even if you don't remember it right now. This way I'll be able to poke you about it in the future and/or perhaps find out what you meant by other menas.

11:05:57 phoe_ s/menas/means/

11:06:29 phoe_ additionally

11:06:45 phoe_ the side effect of me editing the specification is - I've finally learned how to DEFINE-MODIFY-MACRO properly

11:07:25 phoe_ I've always used DEFUN SETF until now

11:08:02 flip214 and is that bad, using DEFUN SETF?

11:08:13 phoe_ no, absolutely

11:08:20 phoe_ DEFUN SETF works

11:08:41 phoe_ it's just that Common Lisp has DEFUN SETF, DEFSETF, DEFINE-MODIFY-MACRO

11:08:54 flip214 "among others"

11:09:00 phoe_ yes, was about to say that

11:09:05 phoe_ oh right

11:09:10 phoe_ SETF FDEFINITION

11:09:23 jackdaniel phoe_: you may add note regarding defsetf, that while it's not required by the spec, many implementaitons allow defining defsetf for multiple-values

11:10:28 phoe_ jackdaniel: added

11:10:31 phoe_ thank you for your input <3

11:10:55 phoe_ oh gods

11:11:05 phoe_ flip214: and there's also DEFSETF short form and DEFSETF long form

11:11:27 phoe_ you stare at that DEFSETF you encounter in your code and it laughs at you, "HAHA THIS IS NOT EVEN MY FINAL FORM"

11:12:40 jackdaniel (setf (point x) (values 1 2)) ; <-

11:13:34 phoe_ jackdaniel: can you write a full example of this behaviour?

11:13:56 phoe_ like, something I can copypaste into the spec as an actual example under Editor Notes?

11:23:21 phoe_ sweet Jesus, some of the examples are really terrible

11:23:26 phoe_ (unless (null x) (rplaca (nthcdr (truncate (1- (list-length x)) 2) x) v))

11:23:49 phoe_ RPLACA NTHCDR TRUNCATE 1- LIST-LENGTH. Just what the fsck.

11:28:15 ogamita phoe_: (when x (setf (car (nthcdr (1- (truncate (length x) 2)))) v))

11:28:26 ogamita + x

11:28:36 ogamita (when x (setf (car (nthcdr (1- (truncate (length x) 2)) x)) v))

11:30:02 ogamita (let ((x (list 1 2 3 4 5)) (v 0)) (when x (setf (car (nthcdr (truncate (1- (length x)) 2) x)) v)) x) #| --> (1 2 0 4 5) |#

11:33:25 jackdaniel phoe_: hum, I was wrong, this is explicitly allowed by the spec. Don't know why I have remembered it that way. There may be any number of store variables.

11:34:19 jackdaniel but you may still use the example if you like it: http://paste.lisp.org/display/338667

11:34:39 jackdaniel scratch the first comment, after double checking it happens I have misremembered

11:37:41 jackdaniel first annotation is better: http://paste.lisp.org/display/338667#1

11:37:54 ogamita jackdaniel: in your example above, you lose a value. (setf (values (point-x p) (point-y p)) (values 1 2))

11:38:10 ogamita I'm not sure you can store multiple values without the (setf values) setter.

11:38:49 jackdaniel I believe it is a conforming code. At first I thought it's not, but I have doulbechecked

11:39:48 jackdaniel (annotation is cleaned up a bit)

11:40:02 scymtym shouldn't it return all values? (as opposed to returning Z)

11:40:48 jackdaniel it returns point not z, but yeah, notes say that it should

11:41:28 scymtym yeah, overlooked the PROG1

11:41:29 ogamita Well, I suppose one could write a setf expander that would allow storing multiple values at once. (setf (point p) (values 1 2)) (assert (and (= 1 (point-x p)) (= 2 (point-y p)))) should be possible (I don't have time to try it out right now).

11:41:45 jackdaniel scymtym: corrected

11:42:54 jackdaniel ogamita: yes, that's what this example is all about

11:46:40 phoe_ jackdaniel:

11:46:42 phoe_ > alexandria:once-only

11:46:59 phoe_ I'm not putting this in the manual for the specification without first defining Alexandria

11:47:40 jackdaniel I've added also x and y to once only, since its in the values

11:47:42 jackdaniel too

11:47:45 phoe_ (which is planned)

11:48:05 phoe_ jackdaniel: yes, but I need an example for standard CL without any external libraries

11:48:25 phoe_ and ONCE-ONLY is an extnernal dependency

11:48:57 phoe_ I mean - I *want* to vote for intertwining Alexandria in the examples at some point because it's pretty much a standard library nowadays

11:48:57 jackdaniel phoe_: then change it to (let ((point point) (x x) …)

11:49:09 phoe_ jackdaniel: okay.

11:49:14 jackdaniel or even better, define once-only, since it's a common idiom

11:49:19 jackdaniel macro should be max a few lines

11:50:54 phoe_ https://github.com/keithj/alexandria/blob/master/macros.lisp#L30

11:50:58 phoe_ >should be max a few lines

11:51:00 phoe_ yeah, right

11:52:10 phoe_ jackdaniel: http://paste.lisp.org/display/338667#4

11:52:13 phoe_ is this correct enough?

11:52:42 jackdaniel yes

11:53:49 scymtym jackdaniel: i think DEFSETF takes care of evaluation order issues: "During the evaluation of the forms, the variables in the lambda-list and the store-variables are bound to names of temporary variables …"

11:54:23 phoe_ okay - one more thing though jackdaniel, I will need explicit output there.

11:54:30 phoe_ to show that the values were actually setf'd.

11:54:38 sjl * wishes alexandria's once-only used the given symbols for the gensym names, instead of naming everything once-only1234

11:54:39 phoe_ so I'll need struct accessors it seems.

11:54:59 phoe_ sjl: make a patch and file it at their mailing list

11:55:11 sjl yeah I should

11:55:12 jackdaniel (let ((point (make-point))) (setf (foo point :kind :2d) (values 1 2)) (print point))

11:56:58 jackdaniel scymtym: you are right

11:57:09 jackdaniel phoe_: there is no need for the once only (nor the let etc)

11:57:25 phoe_ jackdaniel: <3

11:57:48 jackdaniel regarding showing the point, just create it beforehand and print it after defsetfing it

12:00:50 phoe_ jackdaniel: right

12:01:04 jackdaniel here, final version http://paste.lisp.org/display/338667#5

12:01:07 jackdaniel hueh

12:01:49 jackdaniel however the setf could be factored to (setf (values (point-x ,point) (point-y ,point) (point-z ,point)) (values x y z))

12:02:03 jackdaniel so we wouldn't have to put explicit (values x y z) at the end of a progn

12:02:30 phoe_ jackdaniel: not so final then, eh

12:03:20 jackdaniel http://paste.lisp.org/display/338667#6 :)

12:04:17 jackdaniel we could go even futher by doing this in ecase, so setfing 2d point will return two values, but I've got to leave for a while now

12:04:21 jackdaniel :)

12:05:46 jackdaniel I've added it in one more paste, cu

12:13:36 phoe_ jackdaniel: <3

12:13:38 phoe_ clhs defsetf

12:13:39 specbot http://www.lispworks.com/reference/HyperSpec/Body/m_defset.htm

12:13:43 phoe_ the *xy* example

12:13:46 phoe_ it sucks

12:13:49 phoe_ and it sucks *bad*.

12:16:33 Xach Is it an option to skip bad examples for now?

12:21:07 phoe_ Xach: I'm leaving them in for now with a TODO.

12:21:23 phoe_ I only want to remove them only when I have something better.

12:21:27 phoe_ clhs define-setf-expander

12:21:27 specbot http://www.lispworks.com/reference/HyperSpec/Body/m_defi_3.htm

12:21:33 phoe_ In the second example, "Recall that the int form must itself be suitable for SETF."

12:21:39 phoe_ What's the "int form"?

12:22:24 phoe_ I mean, I can see the variable. But why is it named INT?

12:22:33 phoe_ Smells like "integer".

12:23:06 jackdaniel yes, second argument to ldb is integer

12:25:28 phoe_ So what does it mean in this case, be suitable for SETF?

12:25:39 phoe_ That it must be a place?

12:25:47 phoe_ I mean, something SETFable?

12:26:16 jackdaniel http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_p.htm#place

12:26:57 phoe_ jackdaniel: this doesn't answer me yet.

12:27:04 jackdaniel it's just two clicks away

12:27:07 jackdaniel let me click it for you

12:27:08 jackdaniel http://www.lispworks.com/documentation/HyperSpec/Body/05_aa.htm

12:27:16 phoe_ clhs ldb

12:27:16 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_ldb.htm

12:27:45 phoe_ jackdaniel: I'm trying to understand the phrase, "suitable for SETF".

12:28:50 phoe_ From the LDB page, does it mean this? (setf (ldb (byte 2 1) (car a)) 1)

12:29:15 phoe_ So the INTEGER is a place?

12:29:49 sjl phoe_: it's saying that whatever gets given as INT must itself be setfable

12:29:55 jackdaniel (let ((a 3)) a) ; <- a is a place, 3 is not

12:30:13 phoe_ sjl: jackdaniel: I see. Got it.

12:30:30 sjl like you can't do (setf (ldb ... 42) ...)

12:30:41 phoe_ IMO this needs to be said differently, "suitable for SETF" is not a clear term.

12:31:05 phoe_ If I spent five minutes thinking about this and only got a clear answer with your help, other people will get stuck here as well.

12:31:22 phoe_ s/suitable for SETF/a place/ <- better?

12:32:19 sjl I think so, yes

12:34:17 sjl if I were implementing that example I'd say (define-setf-expander ldb (bytespec integer-place &environment env) ...)

12:34:36 sjl to make it clear that a) the second arg must be a place, and b) the place is expected to contain an integer at runtime

12:35:15 sjl Argument Names as a Type System™

12:35:58 jackdaniel glVertex3f

12:36:30 sjl oh yeah that's the perfect storm of using-names-to-compensate

12:49:35 phoe_ sjl: this is actually a sane suggestion, implementing it

13:10:13 pareidolia Does anyone know of a library that does HTTP header parsing like the Accept header in complex cases like this: "*/*;q=0.5, application/xml, application/xhtml+xml, image/png, text/plain;q=0.8, text/html;q=0.9"

13:10:42 ogamita AFAIK, hunchentoot must do it. You can POST stuff to hunchentoot.

13:11:30 ogamita ah, but you mean, for a HTTP client. We use drakma in general. I don't know if drakma manages this.

13:16:14 scymtym pareidolia: more a proof-of-concept but maybe still useful: https://github.com/sbcl/specializable/blob/master/src/accept-specializer/accept-specializer.lisp#L89

13:18:02 axion There is also dexador which seems to split out a hash table of headers

13:18:09 axion spit rather

13:20:10 phoe_ oooh

13:20:13 Trystam ** NICK Tristam

13:20:18 phoe_ I've more or less finished the Flow chapter

13:20:36 phoe_ I'll push it live soon.

13:31:04 Xach i will miss the wordplay of the name hyperspec

13:35:10 pareidolia Xach: wordplay?

13:35:19 flip214 a pune or play on words

13:35:57 Xach the hypertext spec. i guess it's not especially playful, but still.

13:36:09 pareidolia It's more like portmanteau

13:36:15 Xach "the very definition of class" is a bit more wordplay-y, though.

13:36:26 pareidolia Oh yeah

13:36:42 pareidolia I love its timeless design

13:37:35 Xach no need to be cruel :~(

13:37:48 pareidolia No sarcasm!

13:38:08 Xach oh. to me it screams "90s". but i grew up in the 90s, so it has a kind of nostalgic cool. but i wouldn't agree with timeless.

13:38:36 dlowe It's as timeless as your default browser stylesheet

13:39:13 pareidolia Xach: I also love this particular style of typesetting https://image.slidesharecdn.com/borlandcversion3-131119125256-phpapp02/95/borland-c-version30usersguide1991-14-638.jpg?cb=1384865966

13:57:36 phoe_ clhs setq

13:57:36 specbot http://www.lispworks.com/reference/HyperSpec/Body/s_setq.htm

13:57:43 phoe_ why is SETQ a Special Form and not a Special Operator?

13:58:04 beach It *is* a special operator.

13:58:17 phoe_ No! It says Special Form right on the header!

13:58:30 beach It says wrong.

13:58:48 phoe_ Geez, there are so many silly mistakes like this in the spec.

13:58:57 beach special form n. a list, other than a macro form, which is a form with special syntax or special evaluation rules or both, possibly manipulating the evaluation environment or control flow or both. The first element of a special form is a special operator.

13:59:03 phoe_ beach: I'm teasing.

13:59:20 beach Ah.

13:59:25 phoe_ I know it's a special operator, it's just that suddenly, out of nowhere, it says Special Form instead of Special Operator.

13:59:40 phoe_ The specification is full of little things like these.

13:59:47 beach Yeah, the term "special operator" was invented late-ish.

14:01:29 shka well, you can finally set things right

14:01:32 shka to a some degree

14:03:32 jdz Famous last words.

14:09:04 phoe_ set things right

14:10:44 phoe_ (defvar foo 5) (defvar things 'foo) (defvar right 'baz)

14:10:47 phoe_ (set things right)

14:11:06 phoe_ shka: am I doing it correctly?

14:11:20 shka how should i know?

14:15:15 francogrex Hi, is anyone experienced with parenscript? https://common-lisp.net/project/parenscript/

14:15:51 francogrex my question is the following:

14:16:10 francogrex parenscripts translates almost any lisp code

14:16:27 francogrex but the result of the translation may not be a correct JS code?

14:16:52 francogrex it just "javascriptize" the lisp code

14:17:53 francogrex is there a way to make a verification on which code is correct (runs) and which doesn't and where is the problematic translated code?

14:20:24 smokeink_ i'm not sure but i don't think it translates any lisp code, it only translates the javascriptish code

14:23:58 francogrex smokeink_: not really it translates any code and it translates it in any way whether the JS function that is translated exists or not

14:25:08 francogrex see here: http://paste.lisp.org/display/338677

14:25:31 francogrex mapcar, car caddr are meaningless in JS

14:25:41 francogrex yet translation happens

14:26:20 francogrex contrast parenscript with another translator like Linj (lisp to java)... which never translates a code into rubbish

14:28:18 phoe_ francogrex: parenscript does not know what MAPCAR is.

14:28:48 phoe_ read up on its manual - it cannot translate *all* of Common Lisp, and MAPCAR is a part of CL that Parenscript does not know how to translate.

14:30:05 francogrex phoe: you didn't get my point

14:30:22 francogrex i already knew that

14:30:46 francogrex what I am arguing is that it should output anything in such cases...

14:30:49 jurov he wants parenscript to bomb "i cannot translate this"

14:30:56 francogrex exactly!

14:31:28 jurov but it's ambigouos, what if you *want* function named mapcar in js?

14:31:36 ogamita But you can implement mapcar in javascript!

14:31:43 francogrex so right now, compared to http://www.cliki.net/LinJ it is very poor

14:31:57 ogamita The problem might be rather that parenscript doesn't deal with CL packages.

14:31:58 francogrex ogamita: that's not the question here

14:32:22 phoe_ francogrex: uh, I have a different toolchain here

14:32:34 francogrex really?

14:32:37 phoe_ simply let some JS linter process the code that Parenscript spits out.

14:32:46 francogrex try to run it on the code in the link I sent

14:32:46 phoe_ you'll get all undefined vars and functions this way.

14:33:11 francogrex what is it, pls give url of what you have gotten it from

14:33:23 francogrex where

14:34:06 phoe_ francogrex: d'oh

14:34:06 phoe_ https://i.imgtc.com/Yxv2jbP.png

14:34:18 francogrex https://i.imgtc.com/Yxv2jbP.png

14:34:20 phoe_ the first link that I got from "javascript link online"

14:34:45 francogrex ok it's like a debugger

14:34:56 phoe_ yes

14:34:57 francogrex of JS

14:35:04 francogrex to identify bad code

14:35:12 phoe_ and I'd expect you to debug PS-generated code with a JS linter like this.

14:35:23 phoe_ online, offline, anything.

14:35:46 phoe_ there are more sophisticated javascript testing frameworks built on top of NodeJS that I don't know much about, other than the fact they exist.

14:36:11 francogrex ok phoe_

14:36:52 francogrex and how was it again if you want to inject a pure js code into the lisp one before translating with ps

14:36:54 francogrex ?

14:38:41 francogrex (ps (@ document write)) for example you add the @ ?

14:40:16 phoe_ francogrex: I don't know.

14:40:34 phoe_ I've never used Parenscript enough to try that.

14:45:58 francogrex ok

14:49:12 chream Hello! I find myself using adolist and adotimes a lot. Any reason I have never seen them mentioned?

14:49:55 smokeink_ https://www.reddit.com/r/lisp/comments/pb72i/clactivevariables_variables_with_readwrite/

14:50:50 phoe_ chream: what's adolist?

14:51:51 dlowe anaphoric dolist

14:51:58 chream yes

14:52:01 Xach adieulist

14:52:14 dlowe anaphora just isn't that popular

14:52:16 Xach chream: I don't mention them because I think anaphoric macros look bad.

14:53:28 dlowe I used to use acond until I realized I could do (let (target) (cond ((null arg) (reply "no arg")) ((null (setf target (resolve-target arg))) (reply "target not found")) (t ... do something with target ...)))

14:53:31 chream Xach: I agree to some extent, but constantly binding a result var gets frustrating. I would say simple anamorphic macros are ok.

14:53:49 axion I don't agree with temporary symbols being created that weren't requested to be bound

14:53:49 dlowe * is unfrustrated.

14:54:32 phoe_ * published CLUS chapter Flow, it should be online within a few minutes.

14:56:45 chream Xach: might just create a snippet.

14:58:45 jackdaniel chream: it's personal preference. For instance library metabang.bind isn't discussed frequently either

14:59:04 jackdaniel nor some constructs in alexandria – not much to discuss – people aware of this things just use them or not

14:59:11 jackdaniel (methinks)

14:59:13 dlowe bind is an easier sell than anaphoa

15:00:07 chream jackdaniel: cool. havent seen this. Thanks!

15:00:33 phoe_ First preview: http://phoe.tymoon.eu/clus/doku.php?id=clus:todo on the bottom of the page.

15:00:41 phoe_ I'll look more at it in the evening. Pausing for now.

15:01:00 dlowe people do talk about iterate.

15:01:11 phoe_ dlowe: they do, sometimes

15:01:13 phoe_ often enough

15:01:27 dlowe It's funny, though, with all the customization power of CL at hand, I usually just use stuff in the CL package.

15:02:12 dlowe I could use all these cool extensions, but then I'd have to learn them, and then I'd have to integrate them into my practice

15:02:51 dlowe and usually CL has some not-too-odious tool that will do the job already built in

15:04:10 Xach Some people create macros that are so core to what they do that their code seems a bit like another language

15:04:49 Xach ron garrett uses BB a *lot*, for example.

15:08:47 phoe_ What's BB?

15:13:11 chream phoe: https://github.com/rongarret/ergolib. Its short for binding block. replaces let, mvb, db etc. Also does other stuff. I remember one of the first times I tried hacking som lisp library using BB. I could not get anything to work and gave up on lisp for a couple of years.

15:16:32 phoe_ I see.

15:16:45 phoe_ So basically a different metabang.bind.

15:26:37 Xach it is a community of one

15:26:46 Xach like many things in the CL world...

15:27:05 fe[nl]ix I like that

15:27:28 fe[nl]ix it avoid all the political intrigues and drama than happen in big "communities"

15:27:32 fe[nl]ix *avoids

15:27:41 dlowe it avoids all the benefits too

15:27:48 fe[nl]ix true

15:28:06 dlowe no matter how good a singer is, they will not be able to sing a chord

15:28:44 dlowe and sometimes one's ambitions are larger than what one person can accomplish

15:31:44 PuercoPop On small custimization I like is rlet et all. Moving the local function definitions to the bottom improves readability.