libera/commonlisp - IRC Chatlog

17:02:21 skin I have a question regarding unicode support. I know SBCL has it; e.g. `(code-char some-character)` will return that character's unicode code point.

17:02:36 skin (I know about how some graphemes comprise multiple code points, just bear with me)

17:02:48 skin What other implementations are known to support unicode in this way?

17:04:54 beach Did you mean CHAR-CODE?

17:04:59 skin yeah, that one

17:05:02 skin sorry I mix them up

17:05:38 skin How does modern CL deal with unicode?

17:10:35 beach My guess is "not very well". But there is a cl-unicode library by Edi Weitz that some might use.

17:22:07 Shinmera most every implementation uses unicode and utf32 for strings

17:22:20 Shinmera some exceptions exist like abcl (utf16 due to java)

17:24:39 skin How would you convert to utf-8 and back again? Say if you wanted to write to a file

17:25:01 Shinmera most implementations use utf-8 for the external format when writing characters to file.

17:25:09 skin Good to hear. I wondered if that were true, after reading https://github.com/sabracrolleton/uax-15/blob/master/src/trivial-utf-16.lisp

17:25:20 skin OK. So I can basically count on this, more or less.

17:25:29 skin Helpful thanks

17:26:18 Shinmera doing anything else would be very foolhardy

19:02:13 aeth unfortunately, last I checked, most unicode-specific functionality seems to be available in SBCL as sb-unicode and not available portably, while most "unicode" libraries are more about various encodings rather than things like sb-unicode:casefold... perhaps other implementations have similar functionality, in which case a portability library is needed

19:02:41 aeth I notice the lack of any "unicode" on https://portability.cl/ still

19:21:43 Bubblegumdrop Oh man, I always mix up char-code and code-char. Is there a better way to remember which is which?

19:21:48 Bubblegumdrop char->code and code->char ?

19:26:44 ajoberstar Bubblegumdrop: I guess in a OO sense it's kinda like char-code = char.code()

19:27:39 Bubblegumdrop that's kinda helpful

19:30:53 aeth it is basically char->code and code->char yes

19:31:11 aeth in fact, those names would be better so I could imagine an alternate universe where those were the names

19:31:30 aeth though I don't think those -> names exist in CL, while iirc they do in Scheme

19:32:31 Bubblegumdrop I usually carry around my own toolkit.lisp from project to project. That's going in there for sure.

19:33:32 Bubblegumdrop Maybe It's time I made my own toolkit package?

19:33:57 Bubblegumdrop etw.cl the extensible toolkit framework

19:36:49 aeth if you're just going to rename something, don't forget to inline it... there are also other ways to rename something, such as e.g. (setf (fdefinition 'add) #'+) but that seems hackier than having a one-line inline function, where the main disadvantage is that e.g. SBCL by default has an *inline-expansion-limit* of 50 and you can definitely hit that because it's 50 for an entire function

19:40:14 Bubblegumdrop cries in only-once

19:44:21 Bubblegumdrop https://gigamonkeys.com/book/macros-defining-your-own.html

19:44:27 Bubblegumdrop There was a discussion about macros yesterday, let's see

19:44:36 Bubblegumdrop Kingsy https://gigamonkeys.com/book/macros-defining-your-own.html

19:45:02 Bubblegumdrop my copy of PCL should be in my mailbox, let me go check :)

19:45:25 _death using your own names for CL operators is digging an esoteric hole for yourself, and perhaps later on asking everyone who reads your code to come along.. similar to having your own similar-but-different operators.. basically inventing a private language. it's a rite of passage, every lisper goes through it, but abstractions have a cost, so maybe it's better to limit the language invention to your actual application's domain

19:46:46 mfiano I was bored today, so I decided to really study the simple and classic Norvig cons-based queue algorithm. I've been using it for a while in various codes, but haven't sat down to study how it works in detail in some years. I figured out the symmetric analog of storing the tail in the car of the cons, and discovered that it costs an extra traversal for deletion and obtaining the queued elements,

19:46:48 mfiano and removes a traversal from insertion. I'm not too sure if I care. It was just a fun 10 minute experiment. Though interesting that insertion becomes less expensive.

19:51:34 Bubblegumdrop https://3e8.org/pub/scheme/doc/lisp-pointers/v4i4/p2-norvig.pdf ?

19:51:43 Bubblegumdrop I ordered a paperback but they sent me a hardcover! happy day.

19:51:56 Bubblegumdrop Thanks SFPL!

19:55:16 mfiano I suppose I should read the subject matter more fully instead of going by my notes and comments.

19:56:40 mfiano He explored this as well, in his last example of removing the boundary test, and noticed the extra traversal. If the cons queue is reversed, that extra traversal is removed and put in another place, deletion.

19:57:00 Bubblegumdrop oh the reverse reverse

19:57:14 Bubblegumdrop having to explain that to people who don't understand cons...

19:57:20 Bubblegumdrop "why is it backwards?" - "it isn't..."

20:00:24 mfiano it should be noted that insertion and deletion have to be measured for your domain. That is, if you can't afford a vector implementation. It should also be noted that his queue-elements and dequeue functions both increased in traversal with the reverse, but insertion decreases.

20:00:57 mfiano Let's see if I can get the bot to play

20:01:26 Bubblegumdrop I think I did try using vectors but had some esoteric use case that they didn't support, I don't remember why I chose to use plain cons instead

20:01:37 Bubblegumdrop I think I tried arrays too? I actually think I just literally did not understand them at the time.

20:01:40 Bubblegumdrop Actually, ...

20:01:50 Bubblegumdrop https://lispcookbook.github.io/cl-cookbook/arrays.html Thanks!

20:02:26 Bubblegumdrop > A good place to start is Chapter 11 of PCL

20:02:30 Bubblegumdrop :awesome:

20:03:23 Bubblegumdrop I love my lisp books. I think I'm up to 4-5

20:05:14 mfiano ,(let* ((l (list nil)) (q (cons l l))) (setf (cdr q) (setf (cddr q) (list 1))) (setf (cdr q) (setf (cddr q) (list 2))) (setf (car q) (cdar q)) (setf (cdr q) (setf (cddr q) (list 3))) (cdar q))

20:05:21 mfiano ,(let* ((l (list nil)) (q (cons l l))) (setf (car q) (setf (cdar q) (list 1))) (setf (car q) (setf (cdar q) (list 2))) (setf (cdr q) (cddr q)) (setf (car q) (setf (cdar q) (list 3))) (cddr q))

20:06:06 mfiano The latter would just be the inverse of his "removing the boundaries" implementation, with the noted advantage/disadvantages

20:08:53 mfiano Bubblegumdrop: a vector is an array

20:10:31 Bubblegumdrop indeed

20:10:39 Bubblegumdrop that was part of the struggle

20:10:43 Bubblegumdrop I'm kinda new to lisp.

20:11:27 mfiano Yes, it's just that not all arrays are vectors, and not all vectors are simple-arrays, and not all non-simple-array vectors are not simple-vectors....

20:11:32 mfiano I can see the confusion

20:11:47 Bubblegumdrop That was the confusion indeed.

20:11:58 Bubblegumdrop I should make an effort spending some time with arrays.

20:12:12 Bubblegumdrop This page on the cookbook is good and I'm looking forward to reading PCL.

20:13:02 Bubblegumdrop https://github.com/Lisp-Stat/ oh this is cool. So many incredible resources for lisp.

20:18:51 lurkur ** NICK deltaboy

20:54:44 Kingsy Is anyone around that can give me tips on how to handle errors properly in hunchentoot?

21:15:55 skin With regard to https://edicl.github.io/hunchentoot/#conditions ?

21:19:36 Kingsy skin: I have looked over this a few times, but it still doesnt really answer my question. for example. I want to return JSON objects not templates. I also want to be able to specify the object that is returned based on the condition that is raised. I don't see how that is done with the items listed here.

21:22:15 Kingsy the docs are vague. I can see there is a hunchentoot-error class but there is absoutely no information on how it works. What can be overriden and how? etc etc.

21:43:48 Bubblegumdrop Kingsy if you want to return json you may need to set the content type or something, caveman has this

21:44:08 Bubblegumdrop https://github.com/fukamachi/caveman/blob/master/v2/skeleton/src/view.lisp#L33

21:44:24 Bubblegumdrop hunchentoot-error is the base class and you can make your own conditions like so"

21:44:30 Bubblegumdrop (define-condition my-condition (hunchentoot-condition) ...)

21:44:32 Bubblegumdrop I believe.

21:44:59 Bubblegumdrop (hunchentoot:define-easy-handler (say-yo :uri "/yo") (name) (setf (hunchentoot:content-type*) "text/plain") (format nil "Hey~@[ ~A~]!" name))

21:45:02 Bubblegumdrop This is in the docs.

21:45:11 Bubblegumdrop (setf (hunchentoot:content-type*) "application/json")

21:45:15 Bubblegumdrop Should get you what you want.

21:46:44 Bubblegumdrop Kingsy I struggled with the condition system at first too, it's not specific to hunchentoot although hunchentoot may hook into it specifically

21:46:48 Kingsy no I understand that. but its more core than that.

21:46:54 Bubblegumdrop https://lispcookbook.github.io/cl-cookbook/error_handling.html

21:47:09 Bubblegumdrop Yeah you were trying to do something like "catch the error and return json if so"

21:47:18 Kingsy if something goes wrong, in general, it returns an error template for the 500.. I want to change that. so it never returns HTML. ever

21:47:23 Bubblegumdrop If you're able to explain what you want/are trying to acheive in english it's helpful as well

21:47:30 Bubblegumdrop Yes that can be done

21:47:48 Bubblegumdrop I will get back to you if I figure out.

21:48:14 Bubblegumdrop (I did this using caveman

21:48:15 Bubblegumdrop )

21:48:21 Kingsy so lets start with the simple example. I want to. catch ANY error that is thrown during the execution of a route, and return a custom json object, set the content type as you showed. but also log the stack trace to the repl

21:48:23 Bubblegumdrop well, caveman just uses hunchentoot underneath

21:48:26 Bubblegumdrop or woo

21:48:28 Bubblegumdrop or whatever honestly

21:48:45 Kingsy yeah this is just a learning exercise. I know there are things out there that would make this a lot easier

21:48:47 Bubblegumdrop mm yeah so you'll have to wrap all that in a progn if you're using your lambda thingy with ,@body

21:48:56 Bubblegumdrop if you want to evaluate more than one form you should use progn

21:49:06 Bubblegumdrop clhs progn

21:49:07 specbot http://www.lispworks.com/reference/HyperSpec/Body/s_progn.htm

21:49:35 Bubblegumdrop You should be able to just add (log-whatever ..) in the progn and it should just work

21:49:46 Bubblegumdrop However there may be things like the default stream isn't *standard-output*

21:49:50 Bubblegumdrop I will have to look and see.

21:51:01 Bubblegumdrop give me a moment, I'm preparing dinner. chicken and eggplant parm :)

21:51:11 Kingsy Bubblegumdrop: no I got that working. but it meant it stopped logging to REPL, which didnt help. it meant I didnt know what was going on

21:51:39 Kingsy it also got me thinking about what happens when it is deployed. we probably want different error handling based on production and not. so I was just looking for general advice.

21:51:57 Kingsy oh of course, take your time

21:53:05 Bubblegumdrop Kingsy I actually ended up using Shinmera's VERBOSE library because it's so amazing and versatile. This is what I followed to integrate it into my multithreaded web app: https://www.darkchestnut.com/2021/set-up-verbose-multi-threaded-standalone-applications/

21:54:04 Bubblegumdrop https://git.lain.church/Bubblegumdrop/cl-deck-builder2/src/branch/master/src/main.lisp#L90

21:55:46 Kingsy no this is fune, but it doesnt answer the hunchentoot question. I don't know how to properly catch and return different custom objects based on conditions whilst controlling the REPL. the way I am doing it right now feels wrong

21:56:24 Bubblegumdrop there is a section on logging http://edicl.github.io/hunchentoot/#logging

21:58:32 Bubblegumdrop it looks like by default things go to *standard-error*

21:58:53 Bubblegumdrop I've never used *standard-error* lol

21:59:05 Kingsy I'll just try and play with it. thanks for the info!

21:59:15 Bubblegumdrop Derived acceptor classes can implement methods for the ACCEPTOR-LOG-MESSAGE and ACCEPTOR-LOG-ACCESS generic functions

21:59:36 Bubblegumdrop So I think what you'll want to do is have your own acceptor class that subclasses ACCEPETOR: (defclass my-acceptor (hunchentoot:acceptor) ...)

21:59:45 Bubblegumdrop Then pass that to the thingy instead of the default acceptor or whatever you're using now.

22:00:09 Bubblegumdrop You'll want to specialize on ACCEPTOR-LOG-MESSAGE and etc for that class like so: (defmethod acceptor-log-message :after ((obj my-acceptor) ..) ..)

22:00:17 Bubblegumdrop I think.

22:01:47 Bubblegumdrop You can do things like open a log file and save it with defparameter or defvar, (defparameter *my-log-file* (open ...))

22:01:59 Bubblegumdrop I think you should be able to output to the repl like this as well

22:02:23 Bubblegumdrop I'm not sure if hunchentoot uses threads, but threads make this a whole lot more fun.

22:03:58 Bubblegumdrop and by that I mean, *standard-output* of your REPL is *not* *standard-output* of the thread hunchentoot is executing on

22:07:52 Bubblegumdrop verbose handles all of that transparently

22:15:32 Kingsy well I wrote a simple logger and just outputs with format. so I am going to try and use that. will see how it goes.

22:15:53 Kingsy interesting thoughts around threads. I don't know if thats needed yet

22:16:11 Bubblegumdrop GL!

22:16:24 Bubblegumdrop perhaps look for other applications making use of hunchentoot

22:16:31 Bubblegumdrop I don't know any off the top of my head,

22:16:40 Bubblegumdrop https://gitlab.com/myopenbookstore/openbookstore/ I learned a lot from this repository

22:16:54 Kingsy oh ool

22:16:56 Kingsy cool

22:17:06 Bubblegumdrop https://gitlab.com/myopenbookstore/openbookstore/-/blob/master/run.lisp?ref_type=heads

22:17:37 Bubblegumdrop https://gitlab.com/myopenbookstore/openbookstore/-/blob/master/src/web/web.lisp?ref_type=heads#L29

22:18:28 Bubblegumdrop If you find any other programs using hunchentoot I'm interested in knowing more as well. I primarily use woo.

22:18:47 Bubblegumdrop I'm actually not sure what the differences between clack/lack/hunchentoot/woo/etc are

22:19:03 Bubblegumdrop like superficially I understand they're different projects with different goals

0:25:06 paulapatience Shinmera: It is perfectly harmless, but I assume the asterisk is de trop here: https://github.com/Shirakumo/trial/blob/b5e95f859e82faac2db4eb6eba5be26dd1ec80c9/asset.lisp#L48

2:24:26 Bubblegumdrop paulapatience I'm not certain but I think that's an idiom, like LIST and LIST*

2:24:29 Bubblegumdrop clhs list*

2:24:29 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_list_.htm

2:25:02 Bubblegumdrop In this case it does appear to be a typo, though

2:25:09 Bubblegumdrop https://github.com/Shirakumo/trial/blob/b5e95f859e82faac2db4eb6eba5be26dd1ec80c9/asset.lisp#L128

2:25:14 Bubblegumdrop Further down

2:25:42 paulapatience Indeed, I checked that

2:44:37 holycow interesting: https://www.youtube.com/watch?v=fytGL8vzGeQ

2:45:35 holycow that video points to a direction in which lisp programmers might look to find full time positions

2:46:03 holycow it could get interesting, nano tech is coming soon and it will merge with the the LLM stuff happening now

2:53:48 Bubblegumdrop It's Cando? https://github.com/cando-developers/cando

2:54:19 Bubblegumdrop It seems to be associated with Clasp as well

2:54:44 Bubblegumdrop interesting, clasp uses llvm to interop with C++ code

2:55:26 Bubblegumdrop thanks for the linkn holycow

2:55:32 holycow no probs at all

2:55:38 holycow just ran into it my self

2:55:55 holycow this is obviously the next big thing for the next 20 years imho

2:56:09 holycow <-- not associated with this in any way, just adding my opinion here

3:06:53 Bubblegumdrop Clasp appears to be its own implementation of lisp?

3:07:04 Bubblegumdrop Clasp is an implementation of Common Lisp primarily designed for compatibility with C++-language programs and libraries.

3:07:08 Bubblegumdrop interesting.

3:07:32 Bubblegumdrop Oh, theyre in #clasp

3:07:45 Bubblegumdrop oh, drmeister_ appears to be here.

3:07:47 Bubblegumdrop very cool.

3:07:53 holycow oh!

3:07:56 holycow is that who it is?

3:07:58 Bubblegumdrop indeed.

3:08:00 holycow haha!

3:08:22 Bubblegumdrop This is probably the coolest chat I've ever been on. Y'all are brilliant.

3:08:33 drmeister_ Hello

3:08:52 holycow hi!

3:09:03 holycow drmeister_: enjoying watching that video on many levels

3:09:21 holycow especially watching you desribe the beginning of a new industry

3:09:25 drmeister_ Thank you.

3:11:44 drmeister_ I don't have much to say at the moment. I'm working on a proposal.

3:12:28 holycow no probs. if you are open for questions i may have some over the next few weeks

3:15:53 Bubblegumdrop Hah, watching this video is a kick. You went from Smalltalk to lisp. I'm doing the opposite! This is so cool. Thanks for all your hard work drmeister_.

3:16:52 drmeister_ Sure.

3:17:02 drmeister_ We have a new website going live in a week or so.

3:17:45 holycow schafmesitergroup.com?

3:20:48 drmeister_ No, my company.

3:21:02 holycow link?

3:21:23 Bubblegumdrop oh, #ecl is on here too, sweet

3:21:27 drmeister_ Not yet. It's password protected for the time being.

3:21:53 holycow no probs. i don't know what the rules are here but maybe announce it here and on reddit?

3:23:43 bjorkintosh ... and the orange site. Tradition.

3:23:54 drmeister_ It will be here... https://thirdlawtx.com/

3:24:09 drmeister_ "orange site"?

3:24:12 holycow lol @ ornage site. i started to see people call it that recently

3:24:15 holycow hacker news

3:24:52 holycow aha! bookmarked the url drmeister_ thx.

3:25:16 drmeister_ These aren't quite the best audience for our tech.

3:25:36 hayley drmeister_: Did you outgrow thirdlaw.tech?

3:25:39 bjorkintosh they'll find it highly entertaining though :-)

3:27:38 hayley Bubblegumdrop: To misquote Steele, I am now trying to take Lisp programmers halfway to Newspeak.

3:28:00 Bubblegumdrop Lisptalk?

3:28:09 Bubblegumdrop My parents sent me to speech therapy because I talked with a lisp :P

3:28:51 bjorkintosh why didn't they give you ()s instead?

3:29:26 Bubblegumdrop they were BASIC.

3:29:49 bjorkintosh Parens! they mean well.

3:29:51 hayley Bubblegumdrop: Gilad Bracha in turn took Smalltalk programmers halfway to BETA and E.

3:30:10 Bubblegumdrop hm, where do I know that name from

3:30:29 Bubblegumdrop oh, literally.

3:30:52 Bubblegumdrop Totally off topic for this channel. I've been studying Smalltalk. Currently looking at GT and Pharo.

3:31:10 hayley Smalltalk comes up frequently enough in #lispcafe, much to the dismay of gilberth.

4:14:22 beach Why is that a problem for him?