libera/#commonlisp - IRC Chatlog

1:47:23 jcowan I'm just curious, and not much, respectively

1:47:48 jcowan If there were such direct instances of T, the CLOS structure would be somewhat leaky.

1:48:06 pillton jcowan pfd: I recommend the book, The Art of the Metaobject Protocol. It covers the design of CLOS and offers a number other nuggets which are useful for designing/structuring APIs.

1:48:17 jcowan I've read it, yes.

1:48:24 jcowan Though not for some years.

1:51:12 pillton I don't know what "leaky" means, but CLOS doesn't permeate through the language. For example, you cannot inherit from a system class.

1:51:23 pfd Same here. I need to re-read that seminal book. I've been leaning too heavily on the basics _Understanding CLOS_ Lawless/Miller book, mentioned above, which I was fortunate to scoop for not too dear a few years ago.

1:52:20 pillton Even conditions aren't defined to be compatible with CLOS.

1:53:42 jcowan pillton: Sure, that's just a restriction on what the user can do.

1:55:36 Guest74 I don't even know how to begin to debug this. https://imgur.com/a/YSSKrPk  I think I'll just start selling them.

1:57:00 jcowan However, CONDITION is a class in the class hierarchy: a condition object is of class CONDITION.

2:02:56 pillton jcowan: Sure, but it isn't required to be an instance of STANDARD-OBJECT nor any of the standard meta objects.

2:03:29 pillton clhs class

2:03:29 specbot http://www.lispworks.com/reference/HyperSpec/Body/t_class.htm

2:03:44 pillton clhs 7.4.1

2:03:44 specbot Standard Meta-objects: http://www.lispworks.com/reference/HyperSpec/Body/07_da.htm

2:07:25 pillton Actually, I might have to check my definition of the CLOS. I might be using it incorrectly. My definition of the CLOS is as defined in Chapter 7 of the standard. I am not sure if that is correct.

2:10:09 jcowan In what sense can a statement in the standard be said to be "incorrect" (outside errata)?

3:00:11 beach Good morning everyone!

3:02:12 beach pfd: If you have trouble bootstrapping SICL, you can ask questions in #sicl. There shouldn't be any particular problems, but I forgot to push a change to Trucler for a few days and that broke things.

3:03:25 pfd Thanks :beach!

3:03:30 beach Sure.

3:05:15 pfd Oh, OK. :beach, are you Sir Robert Strandh?  Or, may you be another main contributor?

3:06:04 beach pfd: "beach" is the literal translation of "strand" into English, so yes.

3:06:46 pfd :beach Right. Understood Sir!

3:07:18 beach I give you permission to be less formal. :)

3:08:00 pfd Right-o. Thanks :beach.

3:08:28 beach Sure.

3:25:30 contrapunctus pfd: might you be another Indian Lisper? (Hello from Delhi 🙂)

3:27:55 pfd No. I love Indian culture, and I agree with the Russia-India cooperation around energy and ruble-rupee workaround!

3:27:56 pfd I'm Greek/Hellenic born in Germany, and brought up in Toronto, Canada, where I still am.

3:28:56 pfd :contrapunctus sorry I forgot to tag u.

3:38:13 taiju` ** NICK taiju

4:51:14 char[m] Is there some library that facilitates immutable clos? I'm looking for something that will allow me to make a copy of an object while modifying some of the slots of the copy.

5:22:47 Josh_2 Good morning

5:23:31 Josh_2 Can anyone help me out with my macro defining macro, I am trying to pass a class name into the body of the second macro but if I use ,,<class var> I end up with ,<class>

5:25:33 Josh_2 https://plaster.tymoon.eu/view/3038#3038

5:26:26 Josh_2 I edited that paste to include an example and an expansion

5:34:28 mfiano Lots of very strange misconceptions evident

5:36:32 Josh_2 Cool thats not helpful

5:37:22 mfiano I'm trying to think of where to even begin.

5:38:44 char[m] For anyone else, generic-cl's copy fist the bill pretty well. It would be better if it was automatic though.

5:39:53 mfiano LOCALLY is not serving any purpose here. Why are you generating FUNCALL in a quasiquoted form where the operator is known?

5:42:02 mfiano Why is it that you are calling these slot values. Slots are implementation details that shouldn't be exposed, and you shouldn't be associating an accessor with the concept of a slot value. For example, an AROUND method could easily break that association.

5:42:36 Josh_2 Okay listen

5:42:59 Josh_2 How about you help me with my problem, and then perhaps say "hmm maybe you should give your methods a different name"

5:43:02 mfiano With that said, you shouldn't ever need to ",,". Break it up into functions that generate the code for you.

5:43:24 Josh_2 Alright

5:45:04 pillton I'd probably take a step back and ask more fundamental questions. This looks like the wrong way to implement this.

5:45:16 mfiano Yes, this.

5:51:00 pillton It looks like you're are trying to make it easier to implement a method for VALIDATE-PAYMENT-INTENT-SLOT.

5:52:17 Josh_2 Yes but in this case I would have been better off just copy and pasting :joy:

5:53:49 pillton I am certainly not advocating copying and pasting. I can't see what you're copying and pasting nor do I have a mental model of the abstraction you're trying to create.

5:55:44 mfiano "How about you help me with my problem...": The problem is usually more fundamental than what is asked, and we are all volunteers here. I am not obligated to help any way other than how I know to break down a problem, when I have the time.

5:56:19 mfiano I understand the frustration coming here after trying something that failed, but that is no way to ask for help.

5:58:10 Josh_2 What is the point of bringing up secondary issues first?

6:00:07 pillton I don't have an issue with how you asked your question. Your question and code just highlights bigger issues that you need to pause and think about.

6:01:12 Josh_2 I was just trying to save myself having to write two lines of code :joy:

6:02:32 pillton Well you can approach your problem as a complication I can live with today, or you could use your problem to learn something from others about how to create abstractions.

6:02:52 pillton My apologies: a complication you can live with*

6:07:15 Josh_2 :facepalm:

6:18:16 taiju` ** NICK taiju

7:47:01 lisp123mobile is there a built in / easy way to get how many digits in a positive integer?

7:47:24 lisp123mobile i guess i could convert to string and count the length but that seems expensive

7:47:57 rotateq ohai lisp123mobile

7:48:44 rotateq ah for the digits ...

7:48:53 lisp123mobile rotates: hey o/

7:49:01 jackdaniel how about (ceiling (log number 10))

7:49:22 rotateq an easy way would be via (ceiling (log n 10))

7:49:34 rotateq jackdaniel: hmm you were faster :)

7:49:45 lisp123mobile thanks!!

7:50:05 rotateq so this is nice too if you need it for different integer bases

7:51:15 lisp123mobile so my unique id function is complete - start with universal time and add a global counter to it (adjusting for its number of digits)

7:51:23 rotateq cool

7:51:50 jackdaniel why isn't it just global counter?

7:51:58 jackdaniel (or gensym in a matter of fact)

7:52:12 lisp123mobile persistence across sessions

7:53:39 zeroc00l0 Where does it gather the time data from?

7:54:09 lisp123mobile get-universal-time

7:54:57 lisp123mobile although my idea is slightly flawed

7:55:25 zeroc00l0 What happens if the system clock is set back N seconds?

7:55:55 jackdaniel if you have sessions to be persistent with then you may also store the global counter value

7:56:13 lisp123mobile jack daniel - i wanted to avoid that

7:56:17 jackdaniel or use random uuid (v4? I don't remember)

7:56:27 jackdaniel the chance of collision is close to 0

7:56:36 lisp123mobile uuid.lisp requires ironclad

7:56:38 rotateq you *could* write as another possibility a WITH-FOO macro that starts with (let ((*gensym-counter* 0)) ...). maybe, or I don't understand the problem

7:57:02 rotateq or maybe a bad idea of mine again :)

7:57:06 lisp123mobile want to minimise dependencies in general

7:57:30 lisp123mobile zero - fair point but running on my own computer so not worried about that

7:57:42 rotateq ah yes, ironclad is not so small

7:58:40 beach clhs integer-length

7:58:40 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_intege.htm

7:58:42 zeroc00l0 What happens if the program starts, creates an id, crashes, restarts, creates an id, all within the same unit of the system clock?

7:58:46 beach It is probably faster than LOG.

7:59:15 jackdaniel beach: but its with assumed base 2?

7:59:22 beach But you can scale it.

7:59:25 lisp123mobile beach: thanks!

7:59:36 beach ... and it works for arbitrary integers too. LOG may not.

7:59:37 rotateq beach: Yes I thought on that too and wrote them in query, but it was not directly what he asked for. So your advises to me sink in. :)

7:59:41 jackdaniel will using integer-length /and/ scaling be faster than log?

8:00:15 rotateq jackdaniel: Maybe depends how the impl deals with calculating LOG.

8:00:32 jackdaniel lisp123mobile: if you just need make-v4-uuid then you may

8:00:39 lisp123mobile jack daniel: specifically the problem is i can’t store global id in a bknr data store and then use it in another store object. and i don’t want to write to file or another persistent storage for this

8:00:42 jackdaniel just comment out the ironclad part

8:00:47 beach LOG has to be scaled as well.

8:02:17 lisp123mobile jack daniel: ok will have another look, i wonder how it compares speed wise vs using a combo of time / counter

8:02:45 jackdaniel lisp123mobile: can't you just class-allocate the counter?

8:03:44 lisp123mobile i will google that tonight

8:04:11 jackdaniel for class-allocated slots you need an instance too, but the slot is shared across all instances

8:04:23 jackdaniel so it might not be applicable to your use case

8:04:40 lisp123mobile that’s what i was thinking too when you mentioned it

8:04:41 jackdaniel another option is to define a persistent singleton object

8:05:01 jackdaniel called /global-session/ ,)

8:05:18 lisp123mobile that’s what i did

8:05:45 jackdaniel then do you need anything besides an ordinary counter?

8:05:53 lisp123mobile but it’s annoying with bknr because i need to update it outside of the object creations

8:06:20 lisp123mobile the easiest thing is just to write to file a counter of course

8:06:37 jackdaniel hmm

8:07:32 lisp123mobile uuid is probably the cleanest way, as you say it avoids ironclad for the basic functionality

8:07:57 jackdaniel create a fork uuid<=4 :)

8:08:28 lisp123mobile i should :) unique ids are very important as I realise more and more!

8:18:45 mfiano Shamelesss plug: Fast, MIT-licensed UUIDv4 encoder/decoder: https://github.com/mfiano/common-lisp/tree/master/misc/uuid

8:19:07 jackdaniel good to know, thanks for sharing

8:19:30 jackdaniel (and it is 100% relevant, so not really shameless ,p)

8:21:19 mfiano I should probably remove the dependency on seedable-rng. That isn't very attractive :/

8:21:39 mfiano My personal usage always required deterministic random identifiers.

8:24:04 mfiano jackdaniel: I enjoyed your tixel article that popped up in my inbox.

8:24:45 mfiano Looking forward to more graphics algorithm writeups :)

8:25:07 jackdaniel this serie will be about time and motion

8:25:20 jackdaniel (and time travel(!)) ;)

8:25:27 jackdaniel thanks, I'm glad that you've enjoyed it

8:25:57 jackdaniel (feedback was very minute compared to amount of fun I had with it)

8:49:44 rotateq I bet this could help me sometime too seeing how to write such.

8:55:24 zacque Is there a libc cffi library for all the macro defined constants, typedefs and such?

8:56:42 jackdaniel maybe autowrap?

8:57:01 jackdaniel it can at least grok structures and enums

8:57:07 jackdaniel cl-autowrap*

8:58:47 semz_ ** NICK semz

8:59:23 jackdaniel apparently c2ffi has a (limited) support for defines too

8:59:57 jackdaniel (and that's what cl-autowrap uses)

9:04:37 zacque Hmmm... I prefer a libc wrapper library where a client can simply load and use

9:06:03 jackdaniel macro defined constants and typedefs are gone by that time

9:06:07 zacque Very much like: https://docs.rs/libc/latest/libc/

9:06:27 zacque Ya, that's why the library needs to keep track of it

9:06:53 jackdaniel so that's what cl-autowrap does, it groks the .h file to keep track of it

9:08:03 zacque Does the library user need to depend on cl-autowrap?

9:08:15 zacque I'm looking into it right now

9:09:04 zacque Maybe I can use it to autowrap all C headers

9:09:17 lisp123 mfiano: thanks for sharing

9:09:44 jackdaniel that is my impression - it generates the spec file that may be distributed with the project

9:09:47 Nilby libc is big and wrapping all of libc in lisp is mostly pointless, since the native CL are usually preferable. usually people pick the few function they need and wrap those.

9:10:41 _death beach: re. LOG I noticed the other day that its implementation in sbcl is weird.. (log n 2) basically translates to (/ (log n) (log 2)) i.e. it evaluates (log 2) each time.. so I resorted to doing (defun lg (x) (* (log x) #.(/ (log 2))))

9:11:09 Nilby there's some things in libc that are hard to duplicate in CL, e.g. setlocale

9:11:28 jackdaniel Nilby: that's when you want to deal with writing bindings, automatic wrappers are a nobrainer solution for trivial-foolib-bindings

9:11:41 jackdaniel (and less errorprone at that)

9:18:45 rotateq maybe I should rename my newer repo for emulating Racket loop types to "trivial-racket-loops" :)

9:20:17 lisp123 rotateq: good branding :)

9:20:19 jackdaniel imo libs and features ported from scheme/racket to cl should be called unhygienic-<something>

9:20:51 jackdaniel a bit mocking, but oh well, the term itself is a mockery

9:21:03 rotateq lisp123: the idea came up here and maybe it can give some writers a better time

9:21:04 lisp123 I wonder who came up with that term

9:21:12 rotateq jackdaniel: oh :)

9:21:37 jackdaniel lisp123: clearly someone who did not like common lisp. to turn the coin, it could be called 'sterilze' as an opposite to 'organic'

9:21:40 rotateq this "How to mock a mockingbird" is rather trivial

9:21:42 rotateq *book

9:23:22 rotateq jackdaniel: and we want to build organisms, not pyramids

9:23:25 lisp123 although 'little schemer' probably made their language less 'industrial'

9:26:49 rotateq https://upload.wikimedia.org/wikipedia/commons/7/7b/Mandelbrot_p1130861.jpg

9:26:52 rotateq oh sry

9:27:05 rotateq wrong channel :( should have gone to #lispcafe

9:28:34 lisp123 jackdaniel: p.s. thanks for your help on bknr nested objects on reddit

9:29:01 lisp123 makes it much smoother now that I can nest objects...such a great datastore it is

9:36:36 zacque Nilby: But if that function is being used again and again across many libraries, won't that cause a lot of code duplication?

9:37:23 zacque I know nothing about `setlocale`, thanks for pointing it out

9:58:39 beach _death: Interesting!

10:11:29 AndrewYu ** NICK Andrew

10:19:18 Andrew ** NICK werdnA

10:19:20 werdnA ** NICK Andrew

10:57:35 mange Hi! Does anyone feel like helping me work out a MOP issue I'm having? I'd like to choose a different effective slot definition class depending on the values of fields in my direct slot definition. Is there a better way to do it than communicating down the stack with a special variable bound in an around method on COMPUTE-EFFECTIVE-SLOT-DEFINITION?

10:58:58 mange I've tried redefining COMPUTE-EFFECTIVE-SLOT-DEFINITION's primary method, which can work but doesn't seem portable. I've also tried using CHANGE-CLASS to change the slot into the type that I want after it's made but SBCL explicitly checks for and rejects that.

11:37:21 Bike yeah, i think amop says you can't change the class of slot definitions

11:37:45 Bike let me see...

11:37:57 Bike mop effective-slot-definition-class

11:37:57 specbot http://metamodular.com/CLOS-MOP/effective-slot-definition-class.html

11:39:18 Bike yeah, tricky

11:43:46 zacque Wow, first timing seeing the CLOS-MOP site, comes in very handy

11:43:59 beach It does.

11:45:29 zacque beach: Thank you! =D

11:46:01 jackdaniel zacque: check out l1sp.org , it is a metasearch for clhs, mop and a few other specifications

11:46:14 beach Sure. When I was implementing CLOS, I was going crazy trying to use the AMOP book. So I created the site to keep my sanity.

11:47:02 jackdaniel beach: did you type things by hand, copy from pdf or you've converted some existing source code to make clos-mop?

11:47:36 beach Good question. I can't remember. I guess I must have typed it.

11:47:44 jackdaniel I see, thanks

11:48:07 beach The existing web-based version was protected, so I couldn't use it.

11:48:20 rotateq And I'm not sure, it was mentioned in the Keene book, when I make a subclass and give additional `:type` option to a slot that comes from the superclass, they are forged together with AND.

11:51:21 zacque jackdaniel: Thanks! The search query is simply enough, i.e. format = "http://l1sp.org/search?q=%s"

11:51:42 zacque Got to come up with a way to browse it from within Emacs ha!

11:51:54 jackdaniel zacque: thank Xach (he created and maintains it)

11:52:40 zacque Thanks Xach!

11:54:05 zacque Maybe I can write a bot for l1sp site? How does specbot work?

11:55:23 zacque I guess it needs to scan through every reply for keyword, when there is a match, it then replies with the matching item

11:55:41 zacque I meant, the matching lookup value

11:57:18 Xach You're welcome