libera/commonlisp - IRC Chatlog

5:53:25 karlosz_ ** NICK karlosz

6:39:08 karlosz_ ** NICK karlosz

6:46:45 karlosz_ ** NICK karlosz

7:26:04 karlosz_ ** NICK karlosz

11:03:16 Simoneee Does the author of Land of Lisp post here?

11:03:26 Simoneee I want to interview him for my dev.to article

11:03:42 beach I don't think he is here, no.

11:03:50 Simoneee ok thank you

11:04:03 beach Are you aware of my partial review of the book?

11:04:16 beach http://metamodular.com/Books/land-of-lisp.html

11:04:17 Simoneee no

11:04:50 beach It is absolutely horrible and should not have been published in its current form.

11:05:08 Simoneee err

11:06:58 beach I see, we didn't want to know that.

11:11:53 jackdaniel that depends what's the goal of the book - if it is entertainment then it surely achieved its goal; the fact that it made common lisp more approachable is just a side benefit

11:35:05 flip214 ** NICK flip768

11:37:35 flip768 ** NICK flip214

12:08:56 luis It's understandable that a book like Land of Lisp isn't very rigorous wrt terminology.

12:09:46 luis it's certainly arguable that using some sort of simplified terminology is desirable even

12:19:04 beach It is not simpler. It is just wrong and ambiguous.

12:19:34 beach CAR is a "command", but CDR is a "function", for instance.

12:21:05 beach And, sure, it is "understandable" since the author probably does not have the training that (say) engineers do.

12:23:16 beach So it's "understandable", but absolutely not "acceptable".

12:29:53 Nilby https://imgur.com/a/RAcESDl ;o

12:31:06 jackdaniel understandable often beats technically correct. for example "symbols in cl are case-insensitive" (the very first remak) - technically wrong, in practice correct and conveys the essence (being technically correct would make much longer description that would only leave the not-trained reader confused)

12:31:26 jackdaniel s/description/explanation/

12:32:06 rotateq The book gave me a good intro back then, but it didn't take long since I started using other sources and got more confident piece by piece.

12:32:36 jackdaniel I'm not aware of any library that would switch the readtable-case to :preserve

12:33:11 jackdaniel "pedagogical lie" of sort

12:33:33 rotateq Yeah, never have seen it myself, but can be useful when reading other things or writing a special DSL.

12:34:20 mfiano Indeed. If such an introductory book started off with the first chapter of CLR, it would dissuade them from continuing quite quickly.

12:34:38 mfiano (CLR starts off by explaining case sensitivity of symbols)

12:40:14 dbotton Ladies and gentlemen: the story you are about to hear is true. Only the terms have been changed to protect the innocent.

12:41:37 dbotton Perhaps that is a better start for LOL.. lol

12:41:58 ck_ "I just look at the pictures"

12:42:05 Nilby i'm biased from all the times i was grievously injured by a wumpus, and mocked by how much i look the caricature of the old lisper

12:43:28 dbotton The audience was of lol was always hobbyists that would go on if got excited to something more

12:57:01 semz "by default, the symbols you enter are converted to upper case first" would be just as understandable without being incorrect imo

12:57:53 semz going into full detail on read-case and print-case at the start is silly obviously since they're barely ever used, but this is a case where the pedagogical lie seems kind of pointless to begin with

12:57:54 jackdaniel that would not be technically correct. 'by default reader upcases the symbol name before interning the symbol'

12:58:38 jackdaniel then, what is interning? why does it upcase the symbol name? what is a difference between a symbol and its name? yada yada ,)

12:58:39 luis semz: you don't want to play the "technically correct" game with a lisp implementor :D

12:58:44 semz sure but that'd be just as understandable since you can't really talk about symbols without interning

12:58:55 jackdaniel sure you can

12:59:01 semz luis: channeling don quixote here ;)

12:59:49 jackdaniel (unless we assume the thesis we argue about, that all description must be technically correct)

13:00:55 semz well i'd expect an intro to symbols to mention what makes them special. I'd have to check if land of lol talks about interning but surely?

13:00:59 semz ...land of lisp*

13:01:18 luis Anyway, I've never read Land of Lisp so I don't know whether the technical incorrectness pays off

13:01:42 jackdaniel when I was learning common lisp I've first picked "gentle introduction", I fell asleep in the middle of the first chapter explaining what makes symbols special ,)

13:02:04 flip214 ACTION is reminded of physical models... the more exact, the harder to understand and predict

13:02:10 jackdaniel then I've started reading ANSI CL (by Graham) and that was a very good introduction book; later I've read PCL that was also excellent

13:02:13 luis Yeah, same. I actually ignored Lisp for a while because of that book.

13:02:21 semz Take in mind that I'm slightly biased - I see plenty of "pedagogical lies" in bad math books which backfire spectacularly

13:03:10 flip214 semz: OTOH, you can't start with the _total_ complexity anywhere - think of CS starting with field-effect transistors and their nonlinearity?

13:03:34 flip214 so there has to be a kind of abstraction level from which you can go up (and, later on, down as well)

13:04:11 semz sure, I'm not arguing to dump the clhs parts verbatim into the intro

13:04:30 semz my point is that in this case you can avoid incorrect things pretty easily

13:04:48 luis I've heard other people enjoyed the Gentle Introduction to Symbolic Computation book, though. lichteblau for instance.

13:05:21 beach semz: Exactly my point. It takes more work of course.

13:05:58 luis Fair enough.

13:06:50 beach I am baffled by the number of people who argue in favor of incorrect and ambiguous terminology, even though it would be quite possible to use correct terminology while keeping things simple.

13:06:58 Guest74 wait, we're supposed to read books to learn lisp?

13:07:52 beach Guest74: No, you can do what you want, but I would like to make sure that those who do read books get quality information.

13:08:51 jackdaniel I'm surpised that lax terminology is enough to dub a book 'horrible' to the extend of being unpublishable - especially a book that was very well received among novice hobby programmers

13:09:06 Guest74 I dunno. Do you give a calculus book to a kindergartner?

13:09:24 rotateq Good that there aren't so much like sand in the sea, plus good ones and timeless, like PAIP or the Keene book.

13:09:24 beach jackdaniel: It is not "lax". It is wrong, and ambiguous for no reason that laziness on the part of the author.

13:09:39 Guest74 I'm not sure any introduction I had to anything, that wasn't boring to tears, was actually factual.

13:09:41 beach no reason other than

13:09:57 Guest74 beach: you're making assumptions.

13:10:40 beach I think I have had enough of this conversation this time around. Have fun!

13:10:55 jackdaniel I stick to my comment about the first "wrongness" pointed out in the book (and I still argue that a technically correct explanation would be harder to a non-trained reader)

13:11:03 jackdaniel thanks, you too

13:13:04 Guest74 I wonder how much of life is that feeling of enjoyment when you've learned the inaccuracies of your introductions to subjects.

13:14:27 Guest74 Anyways, I'd probably argue in favour of a 'incorrect' introductory text that gets people interested to explore than a dry book that dissuades someone from learning by trying to stick every single piece of correct knowledge into a person before they start doing things.

13:14:56 Guest74 I mean, who are we to argue what works for one person or another.  There are different teachers, and there are different students.

13:16:02 jackdaniel sure, I'm certain I would not enjoy reading 'land of lisp' book myself

13:16:31 Guest74 nah, I didn't either.

13:17:34 Nilby i'm somehow still suprised when people think it's their duty, or even possible, to make sure everyone shares their opinion of quality

13:17:36 AadVersteden[m]1 Sholud anyone have intricate knowledge of cl-ppcre: are equivalent solutions expected to yield vastly different speeds or would it be expected that the compiler merges it all? Eg: `([ab])` versus `(a|b)`

13:18:12 Guest74 Though In my youth when I was autistic I very much understood insisting everything must be done in the manner I believed to be correct. Several decades of interacting with normal intelligence humans has very much dissuaded me from that naive notion.

13:18:24 AadVersteden[m]1 I can benchmark this, but my benchmarks would mostly go towards a specific case and I'm curious to know if the whole difference is moot.

13:18:34 jackdaniel AadVersteden[m]1: recently no-defun-allowed wrote a library that beats cl-ppcre by a big margin

13:18:37 jackdaniel let me look for it

13:18:52 Bike i believe ppcre will not attempt to convert a|b into a character range, so yeah, it could be slower

13:19:16 jackdaniel AadVersteden[m]1: https://github.com/telekons/one-more-re-nightmare

13:19:27 jackdaniel (and a post https://applied-langua.ge/projects/one-more-re-nightmare/)

13:19:29 AadVersteden[m]1 jackdaniel: oh... more to check then :-)

13:19:43 AadVersteden[m]1 Bike: Thank you!

13:20:24 mfiano AadVersteden[m]1: Did you retire your other name?

13:21:12 AadVersteden[m]1 mfiano: I'm on Matrix here, joining this channel. I like it but I'm not sure if I can merge both names. Not sure what I look like on IRC now

13:21:19 Bike equivalence of regexes is, let's see, PSPACE-complete, so in general i'd watch out

13:21:19 semz Guest74: There are different notions of quality sure, but "it's all relative so who knows" is just cheap sophistry

13:21:32 Bike though i think that's only in the weird cases that would be hard generally when building an NFA

13:21:42 mfiano AadVersteden[m]1: AadVersteden[m]1

13:22:38 Guest74 semz: good thing I never said something so silly.

13:23:11 madnificent mfiano: That looks quite horrid :/ Might keep using it if people here don't mind.

13:23:12 semz then I'm not sure what your point was with "who are we to argue what works for one person or another.  There are different teachers, and there are different students"

13:23:53 madnificent Bike: The approach used in one-more-re-nightmare might create less context-switches. Inlining things helps the compiler.

13:26:11 Guest74 semz: how can you judge the correctness of something without knowing everything involved?  This does not necessarily imply everything is relative.  It does imply we do not know what is relative and what isn't.  Relative would be a person that needs correct information insisting that correct information always be given.

13:26:43 jackdaniel semz: what a professional would find not acceptable could suit perfectly an amateur, likewise what a professional would find as an acceptable minimum would be too hard for an amateur; from a hindsight many concepts seem clear and should be described as true to the core as possible, but that's only a knowledge bias

13:27:14 Guest74 ^

13:27:28 jackdaniel s/would be too hard/could be too hard/

13:27:41 madnificent Given the structure of the SPARQL EBNF, I guess this approach could be faster here too. Thanks!

13:28:20 rotateq And who understands something can explain it well too, so also a kid might get it confidently after some time. Or do you see it differently?

13:29:05 jackdaniel well, there is a thing called "the curse of knowledge"

13:29:27 madnificent rotateq: I'm not sure you can lump all people under that umbrella. I know people that are perfectly able to use technologies in depth with a good understanding of them, but that wouldn't explain them well to people not in the know.

13:29:45 Guest74 When I teach I usually try and insert some non-detrimental errors.  It gives those students who enjoy pouncing on incorrect information something to live for.

13:29:49 jackdaniel I'd argue that that's a case more often than not (what madnificent said)

13:29:57 rotateq madnificent: Yeah you can't do it too generally of course.

13:30:35 madnificent Not sure if this is the only thing but: Being good at explaining things _also_ requires to know what context the listener starts from. Some people clearly don't have a clue about the rest of the world XD

13:31:20 Guest74 Those Ivory Towers sure have good insulation.

13:31:42 Guest74 especially built by physicists.

13:32:27 mason So, I think it's safe to assume that LoL has higher-quality and more extensive cartoons than any other book out there.

13:33:00 rotateq madnificent: That even more. :) Not often easy here. And no, I'm also near to physics and concrete math, the Feynman view is good.

13:33:00 mason I just re-read the critique, and I'm not sure I'd catch on "command" versus "function" myself. Adaptive processing.

13:33:20 mason If he said "method" that might tickle a bit since it has a more specific meaning.

13:34:01 mason ...or "instruction" which has a more specific meaning in the other direction.

13:34:19 jackdaniel mason: 'function' is a specified term in the ansi standard, while 'command' is a common word

13:34:34 semz Guest74: I'm not sure what you're trying to say with that tbh, it sounds exactly like what I was criticising except shifted up one level on the meta scale. Although I get jackdaniel's point and don't disagree (in the abstract; obv. disagree how it applies here but then we talked about that already), so if that was the point I guess we don't disagree too much.

13:34:34 mason jackdaniel: But it's hard to not know what he means.

13:34:52 jackdaniel I'm not sure whether it is put like "bla bla `command` doobie dooh", but if it is not, then I'd treat it as an English word

13:35:17 Guest74 I thought he used it as in it's the 'command' in the game?

13:35:36 mason jackdaniel: The specific examples of use are here: http://metamodular.com/Books/land-of-lisp.html

13:35:52 mason First example, "the commands you enter need to be structured as forms"

13:35:56 Guest74 you store the command(game use) in the car and the cdr holds the funciton.

13:36:32 jackdaniel mason: I'm talking about typeface used for particular words. i.e in documentation you may CAPTIALIZE forms, or `backtick` them to signify that they are a formal term

13:36:43 Guest74 page number?  Though you really need the context.  If he's talking about a game command, that has nothing to do with lisp

13:37:13 mason Guest74: I think one of the complaints is that he uses the term in a relaxed manner, describing different things.

13:37:13 Guest74 nevermind 36

13:37:26 jackdaniel either way, I don't have anything to add from my side; I'm just sorry that Simoneee got scared and probably will discourage people from joining this channel in the future

13:37:38 Guest74 I'd have to actually read the book, but I seem to recall it's about issuing game commands and little to do with lisp.

13:37:40 rotateq command may be more a CLIM terminology

13:38:10 Guest74 or command like pick up bucket.

13:38:11 mfiano This discussion really went off the rails huh?

13:38:21 mason Oh, that's unfortunate about Simoneee. I didn't see that.

13:38:25 madnificent Hey, that's good to start a new tangent.

13:39:01 madnificent Anyone know of things that visualize Common Lisp code? For instance, something that generates class diagrams or which functions/methods call which other functions/methods.

13:39:36 qhong Is there a util library somewhere that have a multiple-value version of MAPCAR?

13:39:57 Guest74 Sure, there's this 4 decade old library that graphs things out to postscript.

13:40:13 qhong e.g. (mapcar-values (lambda (x) (values x (1+ x))) '(1 2)) => (1 2) (2 3)

13:40:16 madnificent I recently found something like it for Elixir. Because I didn't know too much about Elixir when I got started I noticed a bunch of things that looked like a bad interface. So I guess such tools have value after all. Was curious to run something similar on some of my Lisp code.

13:41:05 rotateq madnificent: You could look for the clouseau inspector or there's one McCLIM example "draggable graph" which shows a class tree. (in default from NUMBER as the root)

13:41:53 madnificent rotateq: thank you :D

13:43:07 rotateq Was useful when I showed a friend how the number classes are connected.

13:43:30 jackdaniel how about using the clim listener? https://imgur.com/yc5H1cf.png

13:44:04 jackdaniel n.b in the "Show" menu there are predefined commands (ha!) that display hierarchies like this one

13:44:15 madnificent That's nice! Wonder if I can just run McCLIM

13:44:25 rotateq ahh, one of my newer favorite CLIM applications, of course

13:44:34 rotateq madnificent: On which platform are you?

13:44:58 jackdaniel with mcclim you can use also output-only backends, like svg or png, disregarding of the platform

13:45:13 madnificent I'm on Linux, using EXWM

13:46:35 rotateq Ah cool, so the main backend for McCLIM is xserver. Try (ql:quickload :mcclim) in your implementation.

13:48:52 Guest74 "A form is simply a list with a special command at the beginning—typically

13:48:52 Guest74 the name of a function." It seems they're using 'command' as in what you want to command the computer to do.  sorry about kicking the horse again.

13:49:55 madnificent clouseau is quite cool at first sight :D

13:50:06 masinter__ looks like it was derived from Interlisp

13:50:44 jackdaniel masinter__: it is scymtym's original work

13:51:23 beach Is one of you talking about the listener and the other about Clouseau? Just checking.

13:51:44 beach I forget who wrote the CLIM listener. Maybe hefner?

13:51:52 jackdaniel yes, hefner

13:55:17 masinter__ not the code, the name "command" and the idea that when you type in, it isn't REPL it's read / interpret / print loop where interpret looks at the first typed on the line and sees if it matches a "command" name

13:56:29 masinter__ and before, the "grapher" function that took a list representation of a graph and gave you an interactive display layed out

13:56:35 beach masinter__: We just implemented the CLIM II specification. It is entirely possible that it was inspired by Interlisp.

13:56:49 beach The listener is not part of the specification, though.

13:57:31 masinter__ i imagine it was a much cleaner design

13:57:32 jackdaniel I think that the term 'command' predatest interlisp too

13:59:13 masinter__ sure, interlisp took it from Tenex OS from PDP-10s; i'm not sure multix had a command loop

13:59:48 masinter__ The 360 OS had JCL

14:00:07 jackdaniel CLIM presentation system was inspired by https://dspace.mit.edu/bitstream/handle/1721.1/41161/AI_WP_219.pdf and this document seems to use the term command

14:00:27 jackdaniel the working paper is from 1981 so it is possible that it got inspired by interlisp indeed

14:03:41 beach masinter__: Multics had an interactive command processor.

14:05:36 Nilby lisp people have a long history of copying stuff. the tree graphing on lispworks looks similar, as well as symbolics and interlisp

14:06:00 Guest74 it's graphing

14:06:42 Guest74 they kinda all look the same. biggest difference being those with and without curves.

14:09:51 Guest74 So I've got a dilemma with type-1 fonts in how do I store the glyph data?  It's rather elegant the way you can call subroutines to draw specific parts of a font, and makes manipulating those parts easier. But it's not a very practical format.

14:10:58 Guest74 Do I convert every glyph and subroutine into a cl-vectors path so it's easy for anyone to deal with as nobody seemed to suggest any other formats?

14:12:51 masinter__ https://helpx.adobe.com/fonts/kb/postscript-type-1-fonts-end-of-support.html

14:14:11 Guest74 and? They still exist.  they're quite elegant, and don't require a ps interpretor.

14:14:43 masinter__ sorry i just came across that

14:15:08 madnificent jackdaniel: do you know by accident if one-more-re-nightmare (what a negative name -_-) is known to work with multi-byte characters? I mean this as a generic question on how stable this solution is expected to be with respect to edge cases.

14:15:10 Guest74 They also have easier to apply hinting that what you'd have to write to hint a ttf or otf.

14:15:57 jackdaniel madnificent: I have no idea, I have not used it myself; I've just remembered reading about it

14:16:31 Guest74 type-1 -> otf is like opengl ->vulkan.

14:17:05 Nilby with postscript, it's typical to turn everything display lists that are just vectors (i.e. curves are rendered to vectors), which can then be rendered as trapazoids meshes. for modern graphics drivers, it would probably end up as triangle strips

14:18:01 madnificent jackdaniel: okay, hope I can make some time for it. I really like the approach to the problem. I'm using this for LL(1) parsing based on a FSA myself.... assuming I remember my code.

14:18:24 Guest74 The plan is to have that conversion possible at the path level.  While this is more about construction of the paths. i.e. do I construct the path or leave the commands to construct the path and reconstruct it on demand.

14:19:26 Guest74 or, have the user deal with constructing the paths.  Which is a hard question since I'm the only user right now.

14:20:42 Guest74 Though I guess if someone writes a vector library that runs on the gpu they could probably want the simple postscript.

14:21:18 Nilby of course it depends on what you're doing with it. unless your making a font editor, usually the goal is to get to rendering quickly, and perhaps to stick in a font cache

14:21:52 Guest74 font editor and inspector are required.

14:22:21 Guest74 the point is probably moot at the font:atlas stage.

14:23:14 Nilby ah, well then you probably want to keep in the abstract way that calls out to letter forms, since that's a big thing about the way it's designed

14:24:06 Nilby i assume you've use fontforge?

14:24:11 Guest74 yes, the generic font interface will hide all the transformations necessary.

14:24:27 Guest74 once, long ago.

14:26:34 Nilby it's very educational to examine all the details, and the differences in the font design. some fonts, despite how they look, are internally designed like crap, other very elgantly

14:26:42 Guest74 Thankfully I managed to find a postscript name to unicode list buried in archive.org.

14:27:50 Guest74 elegance is what I see in the type-1 fonts, or it's just because it's probably how I would define it.

14:28:14 masinter__ ? i didn't think there was a standard postscript name to unicode mapping

14:28:38 Guest74 yes, two of them.

14:28:44 Guest74 for glyph names

14:29:55 Guest74 you wouldn't think it using google, but that's because most of the information has been removed.

14:32:33 masinter__ didn't postscript and type 1 fonts predate unicode?

14:33:12 Guest74 I guess the overhead to generate the path will be immaterial to the rendering when doing text layout in a vector program, and if you're not using font atlases for fonts what are you doing??

14:33:23 Guest74 besides getting beautiful text layout.

14:34:04 Nilby masinter__: yes, they basically had to make up their own encodings for everything

14:37:09 Guest74 I guess I'll keep the data raw...ish, just decoded.  It'll probably make for a better font inspector when you can visually tell which parts of the glyph are common subroutines.

14:38:07 Guest74 double encryption and encoding, they seemed kind of paranoid.

14:39:21 semz ACTION had no idea Postscript fonts had built-in encryption

14:40:00 Nilby Guest74: it's very expensive to create japanese outline font. but yeah, they were too paranoid because it turned out you could program the printer to upload the raw font data to anything

14:41:25 masinter__ different licensing terms -- in the US, there's an exception for raster fonts in the copyright law, but not in Japan

14:42:36 Guest74 lol, didn't know that.  The early days of encryption and protection always seemed to be circumvented by something trivial.

14:43:50 Guest74 I think I'll add a segment browser as well. It's probably education to see the parts that make up a font.

14:45:25 Nilby sometimes fonts have weird hidden easter eggs in characters that aren't used

14:47:28 Guest74 I don't know why, but this Agora font has things like "donot.bleach" "dryclean.petroleumsolventonly"

14:48:23 Nilby Guest74: heh, those are the wierd symbols for clothing labels

14:49:16 Nilby it don't think those are unicode

14:49:43 Guest74 no, that's adobes ps glyph names

14:50:39 Guest74 I'm guessing the real reason they killed type-1 is because the glyph index is only 8 bits and you had to look everything up by ps names.

14:52:16 masinter__ there were two standards: ISO 10646 unicode for character codes and names, and iso 10036 for glyphs

14:52:28 Nilby the real reason is adobe, apple, and microsoft and their battling software monoplies

14:54:04 Guest74 So I'm just storing glyphs like this then. (66 696 :HSBW 476 360 :RMOVE-TO 39 :CALLSUBR -279 -67 :RMOVE-TO 40 :CALLSUBR 419 :VMOVETO 41 :CALLSUBR :ENDCHAR) which seems very unwelcoming.

14:54:52 Nilby Guest74: that's nice and compact, when you have the rendering part

14:58:54 masinter__ https://www.researchgate.net/profile/Larry_Masinter/publication/237527690_Extensions_to_Common_LISP_to_Support_International_Character_Sets/links/0c960535d33199f90d000000/Extensions-to-Common-LISP-to-Support-International-Character-Sets.pdf?origin=publication_list

14:59:08 Guest74 yeah, I guess it'll save memory.

15:07:04 Nilby masinter__: that turned out to be an ahead of it's time proposal, since we can still use most of it now. i'm happy the extended-char got in there, but i'm sad that most implementations still don't use it. an impressive body of work

15:07:17 Guest74 huh, neutered of all abilities to deal with fonts and glyphs.

15:08:51 masinter__ medley uses the Xerox XCCS internally, but we've added Unicode UTF-8 support as an external-format

15:12:01 masinter__ guest174; it was before most people didn't have raster displays so "neutered" => "before"

15:14:47 Guest74 I really wonder where Xach is.  He's the only person I know who has dealt with path building and I'm wondering about how to deal with construction of multiple paths.

15:15:56 Guest74 Right now I just keep track of one path, you close it, you get it, or can see it at any time with vg:show-path.

15:18:37 Guest74 but with fonts, you're usually dealing with multiple paths.  Do I keep a stack? Do I allow access to the stack or just what's at the top?  If I start a new path without having closed the previous path, do I implicitly close it? Do I allow further additions to it when the path above it has been popped?

15:32:57 random-nick

15:59:20 karlosz_ ** NICK karlosz

16:51:14 contrapunctus Is there a tool which warns you about symbols you have imported into a package, but never used?

16:51:39 mfiano Yes

16:52:00 mfiano SBCL's deterministic profiler will list all unused symbols of the packages you ask to be profiled.

16:52:39 contrapunctus I see...