libera/#lisp - IRC Chatlog

11:28:46 Kala_Ghoda my chinese radio

11:44:07 edgar-rft` ** NICK edgar-rft

15:56:50 tjbw ** NICK tjbw_away

16:14:16 alcor` ** NICK alcor

16:57:34 x8dcc Is there a CL function for converting a Lisp expression into a string?

17:00:48 paulapatience princ-to-string?

17:01:52 x8dcc Perfect :)

17:02:13 paulapatience There is also prin1-to-string, depending on what you want

17:02:23 paulapatience prin1 makes it readable, princ not

17:03:11 x8dcc I don't see the difference, but I guess I would want prin1

17:03:37 paulapatience prin1 sets *print-readably* to t

17:03:49 paulapatience For some things it will be the same, for others not

17:03:59 x8dcc Okay, thanks

17:04:11 paulapatience E.g., prin1 on symbols will put the package prefix if necessary, princ not

17:05:02 x8dcc I see. I guess I will notice once I use it for a bit

17:17:02 serafeim is there a way to test a hunchentoot acceptor? for example, i've modified my acceptor so it adds some variables in the session. how can i test that functionality ?

17:22:36 Bubblegumdrop https://edicl.github.io/hunchentoot/#testing

17:23:06 Bubblegumdrop I think you can use drakma from CL, but curl or wget should work too

17:23:12 Bubblegumdrop There's also dexador

17:24:31 serafeim yes i know i could do the http requests by hand however i'd was wondering if there was a better way to test it; i.e with a fake client that would run the tests and have access to *request* and *session* directly

17:25:50 Bubblegumdrop https://github.com/edicl/hunchentoot/blob/master/test/script.lisp That is essentially what the built in test suite does. I don't know about accessing *request* or *ession*

17:26:57 serafeim Bubblegumdrop: yeah i saw that but it ain't what i had in mind :( to understand what I mean: https://docs.djangoproject.com/en/5.1/topics/testing/tools/#the-test-client

17:27:48 serafeim the tests in script.lisp are more similar to integration tests, not unit tests

17:28:23 Bubblegumdrop https://github.com/edicl/hunchentoot/blob/master/test/script-engine.lisp#L161

17:28:38 Bubblegumdrop Looks pretty similar to me.

17:29:31 Bubblegumdrop Have you tried caveman? https://github.com/fukamachi/caveman

17:29:43 Bubblegumdrop It's not a test suite but it's kind of like django for lisp

17:30:02 Bubblegumdrop It can use hunchentoot or woo or some others as a backend.

17:30:26 serafeim Bubblegumdrop: yes i've heard of it but i'd prefer to avoid too much external libs

17:32:45 Bubblegumdrop caveman pulls in quite a few deps

17:33:16 serafeim also hunchentoot seems fine for now

17:34:10 serafeim if i can't find anything better i'll just use the http client to test it

17:37:35 Bubblegumdrop You could always write your own test client.

17:37:48 Bubblegumdrop I may give it a try after our conversation. It seems useful.

17:38:30 serafeim yes it definitely is

17:40:15 serafeim f.e i've got this function that logs in a user https://dpaste.com/EHM8MKZ9N ; there's no real way to test it

17:40:47 serafeim unless i'm missing something

17:41:40 Bubblegumdrop Yes I see you're doing something similar to what I did.

17:41:57 Bubblegumdrop I wrote a macro, WITH-LOGGED-IN-USER that will wrap a BODY

17:42:18 Bubblegumdrop I'm using caveman but perhaps you can translate a similar idea to hunchentoot https://git.lain.church/Bubblegumdrop/cl-deck-builder2/src/branch/master/src/web/toolkit.lisp#L21

17:43:52 serafeim thanks i'll take a peek at it to get ideas

17:44:32 Bubblegumdrop That way, you can just wrap your routes that require a login with WITH-LOGGED-IN-USER

17:44:51 serafeim however i was thinking of using the easyroot-decorators https://github.com/mmontone/easy-routes?tab=readme-ov-file#decorators to ensure that routes would get a logged in user

17:44:53 Bubblegumdrop And if the login was successful, you can expect the appropriate output, or an HTTP redirect to a login page otherwise

17:45:11 Bubblegumdrop Yes essentially the same concept.

17:45:32 Bubblegumdrop A decorator and a macro are a similar concept.

17:45:46 serafeim yes there are a lot of ways to implement

17:46:23 Bubblegumdrop The freedom to do whatever you please however you please to do it can be an obstacle sometimes.

17:46:58 serafeim ha ha yes analysis paralysis

17:47:24 Bubblegumdrop The best course of action in this case is to just mash out random sh*t. Writing helps your brain coalesce ideas into something coherent.

17:47:48 Bubblegumdrop What are your thoughts on ChatGPT? It's quite helpful in situations like this I've found as well.

17:48:02 Lumie Copilot is better overall

17:48:29 Bubblegumdrop Are you using Copilot for lisp?

17:48:29 Lumie Less hallucinations from my experience

17:48:31 Bubblegumdrop I haven't tried yet.

17:48:37 serafeim hm i don't have very good experience with copilot... i'm using claude.ai and chaptgpt both sometimes

17:48:51 Bubblegumdrop I've only recently signed up with Claude so I haven't formed an opinion on it yet

17:49:05 Bubblegumdrop ChatGPT is amazing at writing prose and my lisp comes across like prose so it's been 11/10 IME

17:49:09 Lumie Bubblegumdrop: I don't really use it for Lisp but I have gotten good results for LaTeX

17:49:12 Bubblegumdrop It really struggles with esoteric stuff

17:49:13 serafeim what is *very* useful to me is that I can copy paste a bootstrap 5 html snippet and ask it to return it to me to spinneret. it works perfectly every time and is very helpful

17:49:23 Bubblegumdrop yes that's exactly what I'm using it for as well

17:49:31 Bubblegumdrop Type this 500 lines of boilerplate I don't want to

17:49:47 Bubblegumdrop You can give it a psuedo-code description of what you want and it will do a pretty good job too.

17:49:53 serafeim yes spinneret is great but everything's in html :)

17:50:23 Bubblegumdrop have you tried djula? Its one of caveman's dependencies https://github.com/mmontone/djula

17:50:31 Bubblegumdrop Djula>>>> IMO

17:55:00 serafeim hm well i am using django for like 15 years so i'd wanted something different for templates

18:00:04 Bubblegumdrop probably for the best.

18:36:04 owen https://github.com/garlic0x1/hiccl

18:36:22 owen best html templating because you can compose with functions instead of macros

18:37:18 owen quasiquoted sxml

18:37:19 wasamasa yeah, I don't get why this isn't much more common style

18:37:41 wasamasa like, the case for macros is a bit of convenience and maybe faster templating speed, but when's the last time that was your bottleneck?

19:02:09 kagevf does hiccl support inline JS?

19:02:35 kagevf I definitely like the idea of using functions instead of macros for this.

19:14:18 ruel ** NICK ruhnet

19:14:21 ruhnet ** NICK lvlinux

19:14:26 lvlinux ** NICK ruhnet

19:15:36 owen kagevf yes

19:16:25 owen using (:script (:raw "js-stuff-here")) iirc

19:18:09 Bubblegumdrop I do pretty much the same thing with djula

19:18:17 Bubblegumdrop {% include "js-stuff" %}

19:19:14 Bubblegumdrop https://parenscript.common-lisp.dev/

19:19:17 owen djula is fine i just want to keep it all in lisp when i write html

19:19:43 owen instead of have a separate templating language with its own editor modes

19:20:23 Bubblegumdrop That was a pretty common source of frustration for me. I hadn't figured out a solution yet.

19:20:37 Bubblegumdrop I had a few ideas.

19:21:32 Bubblegumdrop https://40ants.com/reblocks/ I had come across this too.

19:21:47 serafeim i guess the name hiccl is from https://github.com/weavejester/hiccup

19:29:03 Bubblegumdrop Why are lisp books suddenly so expensive

19:33:54 ieure Because they're just so damn good.

19:44:30 paulapatience render-form being unexported implies that you shouldn't use it

19:44:36 paulapatience (In hiccl)

19:44:44 paulapatience That should probably be fixed

19:48:27 serafeim i'm a little troubled with the hunchentoot multi-thread situation. i see that there's a `(defvar-unbound *request*)` in hunchentoot. this global will be set to be the current request. but what would happen if there are multiple threads ? won't their requests be mixed?

19:49:04 serafeim i asked claude.ai and it answered me this: Special variables in Common Lisp: In Common Lisp, special variables (like *request*) have dynamic scope. This means their bindings are thread-local by default. ; is this true ?

19:49:43 robin serafeim, it'll be implementation-dependent, but iirc yes special variables end up being thread-local

19:50:00 serafeim interesting i didn't know that

19:51:06 robin serafeim, i think sbcl for example will "inherit" the initial thread-local value from the creator thread, which may be the reason for "defvar-unbound" (whatever that does -- just a special declamation?)

19:51:46 serafeim it says: "Convenience macro to declare unbound special variables with a documentation string."

19:54:58 Bubblegumdrop Huh, Debian has a package for hyperspec

19:55:00 Bubblegumdrop neato

19:55:12 Bubblegumdrop Thanks whoever mentioned clisp, I never thought I'd be trying different CL implementations

20:22:42 owen paulapatience you should just need to use hiccl:render

20:26:12 owen the only reason a user would touch hiccl::render-form is to specialize on a class or something for a special case

20:34:49 kagevf owen: cool, I will check out hccl more at some point ... I like cl-who, but I can't figure out how to get it handle markup at runtime - such as from user input - without using eval ...

20:36:40 paulapatience owen: Sure, but if the usecase of specializing to a class is meant to be allowed for users, the function should be exported.

20:36:55 paulapatience And I think that functionality is quite useful.

20:37:50 owen okay, i will tell garlic to export render-form and apply-tag, it's a good idea

21:27:10 HER_ ** NICK HER

21:32:18 comatory- ** NICK comatory

22:10:18 levitating__ ** NICK levitating