libera/#commonlisp - IRC Chatlog

4:17:03 yates_work in emacs, you can get help on any elisp function via C-h f <function> RET. is there a similar thing for cl under emacs/SLIME?

4:18:40 yates_work nm

4:18:48 yates_work that was lazy. i'm rtfm

4:18:53 yates_work (the slime manual)

4:29:58 beach yates_work: It is good you didn't go for C++, because not only is the language very complex, it is also impossible to write a large C++ program that is both modular and fast. But I have said this many times, so I probably won't repeat it.

5:24:47 yates_work beach: thanks for your confidence and faith in my abilities..

5:26:07 yates_work ACTION reminds himself that 26 years writing C++ is still the beginner stage..

5:26:58 beach Common Lisp takes a while to learn as well, at least of you want to be good. But it is a much simpler language.

5:36:08 beach yates_work: I am not sure whether that remark of yours was sarcastic, but I have absolutely no doubt about your abilities. But I can't say I have confidence in them either, because I haven't seen much of your work.

5:56:10 elderK I have been summoned by the talk of C++, Common Lisp and Learninations!

5:56:21 elderK But also: Happy Near Year in advance, guys!

5:56:32 elderK May you have many enjoyable days of Lisp ahead in 2024 :)

6:08:23 elderK yates_work: I'm primarily a C/C++programmer, too. I've been playing with CL for many years but haven't really done anything truly serious yet :) I wish you luck in your Lisp journey.

6:08:48 elderK I'm sure you've already seen this but if not, you might find this useful: https://lispcookbook.github.io/cl-cookbook/

6:11:44 beach elderK: What is holding you back from using Common Lisp? Is it just that you need more time?

6:11:59 elderK Hey beach!

6:12:02 beach Hello.

6:12:15 elderK It's time but also having hard time breaking C/C++ preconceptions.

6:12:30 elderK It may also just be that the things I want to work on, are not that good of a match for Common Lisp.

6:12:36 beach I see.

6:12:58 elderK I figure it's inexperience more than anything - there is a lot I still don't know.

6:13:25 beach Sure. Lack of experience is fixable.

6:13:54 elderK For instance, take arrays in CL. In C++, when I have a vector, I can specify whether that vector will contain instances of whatever type by value, or whether it will contain references to instances of whatever type.

6:13:55 beach But I am curious about things that are not a good match for Common Lisp. I can perhaps imagine the lack of certain libraries.

6:14:30 beach But why would you want to do that?

6:15:23 beach I would phrase that as "In C++, when I have a vector, I HAVE TO specify..."

6:16:07 elderK Having to specify sucks but it's useful to be able to control how things are laid out: For efficiency purposes, it is great to have all elements of a vector be allocated in contiguous memory.

6:17:28 beach I can see that, but then I can also see that the overall performance of most applications would suffer from that. Because then, when you want to pass an object to a module, you have to copy it.

6:17:48 elderK That's not necessarily true: You can pass by reference.

6:18:00 beach But then your modularity goes down the drain.

6:18:12 beach Hence "you can't write a program that is both modular and fast".

6:18:31 beach But I fully admit that there are applications where you don't care about modularity.

6:18:39 elderK What do you mean by modularity here? Even in CL, when you pass something, you may not always be passing a deep copy but instead a reference, right?

6:19:16 elderK You have a lot more knowledge than me here, beach so I am very keen to hear what you have to say. Why would passing a reference to some data break modularity?

6:19:18 beach Yes, but then the module and the client of the module must have a very tight relationship about what it does with the objects.

6:19:57 beach But again, some applications don't need modularity.

6:20:10 elderK Perhaps not. Generally if you are passing some instance to another module, by reference, there will be a convention as to how that object is to be used. You might be passing it by const-reference, so that you can query it without performing a copy.

6:20:19 elderK Passing by mutable reference is considered a bad smell mostly.

6:20:45 beach But then you can't have a module that holds on to the object.

6:20:53 elderK No, you still can.

6:21:00 elderK Just because you pass it by reference doesn't mean you cannot copy that.

6:21:06 elderK It's just under your control when that copy is made.

6:21:22 beach Oh, but if you copy, your performance goes down the drain.

6:21:22 beach

6:21:40 beach So either your modularity or your performance goes down the drain.

6:21:51 elderK Yes but it would be the same in CL, right? If we pass a reference to something to some function, that function may or may not mutate it. So, you might pass a copy instead.

6:23:16 beach Well, that *might* be the case. But if a module might mutate an object, it does so because it's the very purpose of that module.

6:23:36 beach So you would not copy an object just to be sure it doesn't get mutated.

6:23:51 elderK Right but in this case, how does that differ from C++?

6:24:09 elderK There, if we had some module that was intended to mutate some data, and we pass it a reference to that data, well, it can mutate it. How is that any different?

6:24:14 elderK The module dosn'

6:24:50 elderK Sorry - ergonomic keyboard. The module doesn't care where the thing lives, as long as it can access and modify it. If it is intended that other modules be able to hold onto that stuff, then that would have to be factored into the original design.

6:24:56 elderK If that's the case here, then I agree, that sucks.

6:26:17 elderK beach: I would love to be able to use CL as my primary language, for all things. Even today, I tell my wife about CL :P

6:26:29 beach In a language without automatic memory management, you must have a much tighter coupling between a module and its clients. If a module doesn't mutate an object, you would either still have to pass it by reference for reasons of performance, or you would have to copy it. In the first case, modularity suffers, and in the second case, performance does.

6:27:23 elderK Right - because the modules have to have some protocol to control when some shared data is released, if you pass things by reference and don't want to copy.

6:27:27 elderK That's fair.

6:28:14 elderK In C++, you can opt to use more GC-like behavior. You could for instance, use shared_ptr or weak_ptr.

6:28:36 elderK Granted, those are reference-counted so probably aren't as efficient or safe as a proper generational GC.

6:28:40 beach Yes, and then your performance goes down the drain.

6:29:09 elderK I do have a question for you - and it might help my misunderstanding of performance cost.

6:30:08 beach A simple assignment like a = b turns into "decrement the reference counter of a (which is a memory operation), check whether it is 0 (and if so, deallocate), increment the reference counter of b (which is another memory operation), and then assign".

6:30:08 elderK So, in C/C++, allocation is seen as something that is very expensive. Most programmers these days don't care so much but people who've been coding for a lot longer may still expect an allocation to be very expensive. Excessive indirection, too. That's why we tend to use stuff like vector<Foo> rather than vector<unique_ptr<Foo>>.

6:30:30 elderK But, in something like Lisp, the GC is very smart: Allocation is cheap because you're probably just bumping a pointer in some nursery or something.

6:30:47 elderK So, even though your vector of objects is all references to instances allocated from the heap, it's probably closer in memory.

6:31:19 elderK beach: Yup.

6:31:31 hayley shared_ptr *is* garbage collection.

6:31:38 elderK hayley: I know. It's RC.

6:32:14 beach That's possible, but not guaranteed. But again, I fully understand that there are applications with extreme performance needs, like keeping things close in memory. I am just saying that then modularity becomes a problem.

6:33:18 hayley https://shipilev.net/jvm/anatomy-quarks/11-moving-gc-locality/

6:33:25 elderK Aye: I see modularity issue now, too. Dealing with those issues is very much part of being a C++ programmer. It becomes kind of a second-nature thing. You're right - it does hurt modularity.

6:33:58 elderK There are a lot of things a C++ programmer will work around without even really thinking that hard about it.

6:34:09 hayley Someone showed me a program which runs 30% slower with my GC, due to worse locality, due to my collector usually not moving objects.

6:34:31 beach elderK: And I claim the maintenance burden will increase as a result.

6:34:40 elderK beach: And you'd be right :)

6:34:56 elderK You see? :P C/C++ preconceptions holding me back.

6:35:01 beach elderK: Take a program such as MuseScore. I see absolutely no reason to write such a thing in C++.

6:35:29 elderK Agreed.

6:36:28 elderK I think hardware is fast enough now that even for stuff that may have once required a language like C or C++, higher-level languages are more than good enough. Say, C# and Unity for tons of games. C# and XNA for others.

6:36:34 beach Oh, and they write it in C++ for speed, but then they stick in a scripting language implemented as an interpreter. So when people write scripts, performance suffers even more.

6:36:53 elderK Heh. That's an excellent point :P

6:37:09 elderK One of my favorite games - Homeworld 2 - does that. Their engine is mostly C++, but then the core gameplay logic is Lua.

6:37:25 beach If it were written in Common Lisp, there would be a single language, and user scripting code could be compiled.

6:37:51 elderK Yup and if you really needed to, you could still implement "super performance conscious" parts in something else and call it from CL.

6:38:12 beach That would be a much better strategy.

6:40:23 elderK Something else I have to get used to, and it may also be a "meh" thing is that in C++, you generally always know how large things are. Say, if your class contains a num_whatever member, you can decide whether that counter will be 8, 16, 32, 64 bits in length. That's not a great example. A better example would be if you're say, implementing an emulator and want to very efficiently access parts of emulated memory or tweak bits of

6:40:23 elderK the emulated machine's registers.

6:40:56 elderK Is it correct that you can do the same in CL, provided you specify the proper type information?

6:42:44 beach Hard to say (for me). But we are again talking about extreme performance, and I totally understand why applications with extreme performance needs must be written in a language where you can guarantee memory layout.

6:43:31 elderK I wish there was like, a strange in-between: Syntax like Lisp but behavior kind of a mix of both C and CL.

6:43:41 beach But things like score editors, word processors, spreadsheets, aren't such applications.

6:43:47 elderK Although I imagine the two approaches are kind of diametrically opposed.

6:43:48 beach Still they are written in C++.

6:44:03 elderK Aye. I see no reason they can't be in CL.

6:44:08 elderK Probably in C++ just because "more mind share."

6:44:36 beach Yes, and that's the reason of the absolute disastrous state of the software industry.

6:44:58 elderK Almost ten years ago, I was working at an eCommerce company. There, I managed to get one of my colleagues interested in CL and Scheme. My boss at the time, however, well, he was less than great. He ridiculed me for my interest in CL, and for teaching my friend.

6:45:31 elderK People have a very strange view of Lisp. Even today, most people don't know that it can be really, really fast. I bet most of the time, SBCL and CL in general are probably just as fast as C# or faster. And yet, C# is everywhere.

6:45:39 beach Yes, that's know as having a "closed mind set". Such people are very dangerous. Check out the writings of Carol Dweck.

6:45:51 elderK Here in NZ, CL jobs are nowhere, C++ jobs are rare and C# jobs are *everywhere*

6:45:57 hayley The hell of systems programming is the systems, not that the systems aren't programmed with prefix notation.

6:46:32 elderK So going to quote you, hayley.

6:46:40 elderK :)

6:46:40 beach elderK: I have probably asked this before, but where in Aotearoa are you located.

6:47:18 elderK NP :) I live in Christchurch, a city in the South Island.

6:47:53 beach elderK: I see. When I spent my year in Auckland, I found that there were around 5 Lispers within a radius of a 3 hour flight.

6:48:28 beach In Europe, there are hundreds.

6:48:35 elderK beach: Wow. I wish I could've found so many. When I was at University in Dunedin, people were not very interested in Lisp. Some of the professors liked Lisp but no students.

6:49:00 hayley elderK: Or quote Bob Barton: "Systems programmers are the high priests of a low cult."

6:49:07 elderK Aye :) I remembering running into some French people while on a bus ride between cities: They knew I was writing Scheme. They were AI researchers, I think. Nice people.

6:50:09 elderK beach: When did you spend your year in Auckland?

6:50:31 elderK I lived there around 2014-2015, moved back to Dunedin in 2016 and to Christchurch in 2019.

6:50:51 beach I would have to check. A long time ago now. Probably more than 15 years ago.

6:51:07 elderK Wow :) I wonder how much different it was back then.

6:51:25 beach I am guessing "very".

6:51:31 elderK :P I have been collecting old machines, mostly because I'm curious how different writing software was back in the 80s or early 90s, from now.

6:52:44 beach 2005-2006 it seems.

6:53:29 elderK Wow. Maybe there are like, Lisp groups around here that I haven't found.

6:53:47 elderK It'd be really neat to ask and learn from people like, in the flesh.

6:53:52 beach I started the NZCLUG then.

6:53:58 beach But it didn't survive my leaving.

6:54:19 elderK :( That'll explain why Google had no results.

6:54:21 elderK That really sucks :(

6:54:24 beach That would be New Zealand Common Lisp User Group.

7:00:14 elderK :) I will give CL another shot.

7:00:42 elderK Even though I don't use it often, or for anything real, I still enjoy reading about it a lot and I would like to use it for real stuff.

7:06:29 beach elderK: Perhaps you could pick up one of the projects I list on my "suggested projects" page.

7:07:18 beach http://metamodulaire.org/Common-Lisp/suggested-projects.html

7:07:41 beach Some of them are being worked on, so consult us before taking something on.

7:08:01 elderK The assembler seems interesting.

7:08:25 elderK The last remaining issue is simply time: These days, distraction-free time for programming is very rare.

7:08:34 elderK Unless, of course, it is for work :(

7:08:42 beach We have a partial assembler for x86 code.

7:08:57 beach I fully understand.

7:09:37 elderK Is the assembler's source open?

7:10:19 beach Yes https://github.com/robert-strandh/Cluster

7:10:47 elderK Neat :) I'll take a peek.

7:10:56 beach Great!

7:10:59 elderK Hopefully what I learned from LiSP will be helpful :)

7:11:16 beach The book?

7:11:55 beach That's a great book, especially the English version. It's because of the excellent translator.

7:12:03 elderK Aye. There was an exercise there, where a naive interpreter was written. But, instead of just you know, evaluating the source as it went, it translated everything into objects, first. Then executed them.

7:12:14 elderK I love that book. :)

7:12:33 beach The translator is my wife.

7:13:02 beach If you come to ELS and bring your copy, she will sign it for you. :)

7:15:14 elderK :D WOW!

7:15:26 elderK She did a very good job!

7:15:32 beach Yes, indeed.

7:15:51 elderK Please send her my profuse thank yous, as all of that knowledge would've been inaccessible to me without her work.

7:16:02 beach Will do.

7:16:24 elderK :P How much money would it take to translate the latest LisP edition? :P

7:16:26 elderK Heh.

7:16:57 beach Let me put it this way.... The money she made for that translation paid for my private sauna. :)

7:17:02 beach Custom made.

7:18:28 elderK Too bad we couldn't like, get a GoFundMe or Kickstarter going :)

7:19:12 beach You would still have to find a translator. I am pretty sure she wouldn't do it again.

7:19:38 elderK I bet it was a huge job.

7:20:00 elderK My wife is a part-time writer, I can respect how hard it is to write a story, littlelone translate something else.

7:20:11 beach Yeah, I think she went to Paris to see the author pretty much every week for quite a while.

7:21:28 elderK Wow. Did she (you guys?) live in Europe then?

7:22:02 beach Yes, we have lived here since the end of 1987.

7:22:26 elderK Wow :) That was when I was born!

7:22:31 beach Heh.

7:23:37 elderK My wife is from Texas but has a love for Britain and Europe in general. We both hope to visit some time in the future. Whenever we talk about where we will go, I always try to wiggle in a visit to somewhere where a Lisp conference or something might be :)

7:23:52 elderK I'd love to meet you and others here. Also, some of the #chicken Scheme crowd. They are really nice people, too.

7:24:15 beach Oh, my wife is also originally from Texas. Interesting.

7:24:27 beach I say plan to go to ELS 2024.

7:24:41 beach And the preceding SBCL conference too.

7:25:06 elderK Man, what a small world.

7:25:10 elderK :P San Antonio by chance?

7:25:17 beach No, Houston.

7:26:09 elderK That's cool :) I'd like to visit there sometime, too, just for the NASA stuff.

7:26:30 beach She says that Texas is a good place to be FROM.

7:27:09 elderK My wife says the same.

7:27:35 elderK She prefers the colder weather here :)

7:27:35 beach I am serious about the ELS suggestion.

7:27:58 elderK https://european-lisp-symposium.org/2024/ ?

7:28:05 beach Yes.

7:29:03 elderK That'd be awesome although I doubt I'd be able to get that time off work :(

7:29:09 beach :(

7:29:23 beach There are usually more than 75 participants, and sometimes as many as 90.

7:29:57 elderK Man, that's awesome.

7:29:59 elderK Is ELS yearly?

7:30:12 beach It is.

7:30:24 beach In a different European city every year.

7:30:58 elderK Perhaps in a year or so :) I'll bounce the idea to my wife :P

7:31:05 elderK If there's ever one in Ireland, she'd jump for joy.

7:31:39 beach We would need a volunteer to organize it.

7:32:15 beach But if you know someone, then have that person volunteer and it will very likely happen.

7:33:52 elderK I don't know anyone but I will do my best to become friends with an Irish lisper :)

7:34:06 beach Good plan! :)

7:46:56 Guest24 Hi, would any body help me find out why a ironclad rsa signature different from what is signed with openssl? The code is at https://plaster.tymoon.eu/view/4058

7:49:39 Guest24 ** NICK tsanhwa

8:25:44 olnw beach: Is the English title of the "LiSP" book "Lisp in Small Pieces"?

8:27:25 beach Yes.

8:27:53 olnw Great. I'll see if I can find a copy.

8:28:26 beach I think it is still available, but it might be expensive.

8:29:46 beach olnw: It is not for application programmers. It is about how Lisp is implemented.

8:30:33 olnw Well, I am interested in learning about such things.

8:30:45 beach Good. Just wanted to make that clear.

9:26:13 beach olnw: Also, feel free to ask questions here.

10:03:58 dnhester26 olnw: yes, that's exactly right. I didn't even remember anymore and just read the last commit in the source folder which github adds in the top and it's exactly what you said: https://github.com/lisp-docs/cl-standard/tree/master/dpANS3R%2B

10:04:12 dnhester26 olnw: "created dpANSR3+ which is dpANS3 with the dpANSr changes applied for …ease of use... there is nothing new here..."

10:04:23 dnhester26 olnw: will change it now

10:13:28 paulapatience beach: Where can I find more info about the SBCL conference?

10:17:15 beach Not sure, sorry. I am mainly interested in ELS. I think SBCL is Thursday and Friday preceding ELS.

10:18:04 beach May 2-3.

10:19:50 beach It is called SBCL25, presumably referring to the age.

10:20:01 beach But I see no further information about it yet.

10:39:02 olnw beach: Thank you. I have recently had a renewed interest in programming, especially Common Lisp. I hope that when I am more advanced, I can help with some of your suggested projects.

10:40:30 olnw dnhester26: You say "The CL Hyperspec is the only rendering there is based on this document [the ANSI standard] besides ordering a paper copy from ANSI." I don't think that's true, since Allegro also hosts a free online rendition of the standard.

10:44:23 olnw https://franz.com/support/documentation/11.0/introduction.html

10:47:19 paulapatience beach: I see. Unfortunately this year I will be unable to go to ELS, but starting from 2025 I should be able to.

12:05:27 dnhester26 olnw: thanks, I see, they have the link here: https://franz.com/support/documentation/11.0/ansicl/ansicl.htm I didn't know about this one.

12:05:38 dnhester26 olnw: deploying the upadted now, thanks!

12:10:11 olnw dnhester26: No problem. I now see the note about Allegro's documentation, however the sentence I quoted above is still present on the page.

12:29:54 dnhester26 olnw: just realized that I wrote something similar multiple times. Now checking again...

12:31:25 alip ** NICK syd

12:31:33 syd ** NICK sydbot

12:32:41 sydbot ** NICK alip

12:35:18 beach paulapatience: Too bad about 2024. Hope to see you at ELS2025 then.

12:37:53 beach olnw: Great!

12:40:27 dnhester26 olnw: great, should be online now! thanks again! :D

12:40:46 olnw Sure, no problem.

13:04:04 hayley Happy new year!

13:04:46 beach hayley: You too!

13:24:07 dnhester26 happy new year! :D

13:24:49 dnhester26 Does anyone know a clear guide on writing documentation AKA docstrings in CL? I never found one...

13:25:34 beach Do you mean the syntax, where to put them, or what to put in them?

13:25:42 dnhester26 I added this as list of the questions: https://lisp-docs.github.io/docs/tutorial/documentation

13:25:52 dnhester26 yeah, syntax, where to put them mostly

13:26:07 dnhester26 in functions, generics, methods, clos, etc...

13:26:48 beach I prefer to keep them separate from the code, using (SETF DOCUMENTATION) because documentation strings are noise to the person reading the code; they are meant for the user, not the maintainer.

13:27:31 beach But otherwise, you could provide documentation strings for every entity that is part of the protocol.

13:28:15 beach It is a bit silly to provide documentation strings for internal entities since they don't really have a user separate from the maintainer.

13:28:37 beach ACTION vanishes for a while.

13:28:44 jcowan The difficulty with out-of-band doc strings is that they require an extra step by the person writing the code, which has to be enforced by more painful code review.

13:29:49 beach The difficulty with in-band documentation strings is that they represent noise to the person reading the code, thereby discouraging sufficiently good documentation strings in favor of short, useless ones.

13:29:59 beach ... like the one for CAR in SBCL.

13:30:14 jcowan As for silly, the maintainer is not the only user, because maintainers change over time in two senses: a new human being may be the maintainer, or even if the person is the same, the personality changes -- "old men forget", and young ones too, given the distractions of life.

13:31:43 jcowan But I certainly agree that documenting things that don't need documenting merely to meet an arbitrary target *is* silly.

13:32:47 jcowan On C++, note that it is perfectly possible to do C++ with garbage collection. I retrofitted a C interpreter that was riddled with allocation bugs with the BDW GC, and poof, 30% of the code vanished.

13:32:52 jcowan A huge improvement.

13:33:09 jcowan (Granted, I had written an interpreter for the same language in Scheme first.)

13:35:15 jcowan s/a C interpreter/an existing C interpreter written by someone else/

13:35:47 Shinmera People in C++ love to use garbage collection anyway, they just call it "smart pointers" and other silly stuff

13:36:16 jcowan Fair, but in this case I meant conventional non-moving GC rather than reference counting.

13:36:53 jcowan (If you are going to call refcounting "GC" on theoretical grounds, what do you call what the rest of us call "GC"?)

13:37:36 moon-child tracing gc

13:37:46 dnhester26 in-band meaning in the code itself and out-of-band meaning with setf?

13:37:46 jcowan Thanks.

13:37:56 jcowan Yes, at least whne I say it.

13:39:30 dnhester26 thanks

13:40:17 jcowan finally, of course both performance and modularity can be achieved without GC of any sort, if you have enough money, brains, and time to do the necessary manual proofs. After all, humanity got people to the Moon and (more importantly) safely back to Earth with 1960s technology, both hardware and software.

13:40:49 Shinmera they still had a memory leak tho

13:40:52 Shinmera (lol)

13:56:37 beach jcowan: Comments, not documentation strings, are for the maintainer.

13:57:30 beach Remarks meant to clarify to a maintainer how the code is written and why it is written that way, belong in comments.

14:01:24 hayley jcowan: Someone once said that text rendering is harder than getting to the moon and back. I wonder if that's true, and how big the gap is.

14:05:24 beach I once consulted for a company that had written a biggish application in C++ and they had problems both with memory leaks and with freeing live objects. So I suggested the use the Boehm-Demers-Weiser garbage collector. The two project leaders then told me that they thought it was the responsibility of the programmers to "clean up after themselves".

14:05:54 beach I was so dumbfounded I didn't know what to say.

14:05:55 jcowan "Then stop pissing on them", would have been my response.

14:06:33 jcowan Or a softer variant if I thought I could change leaders' minds.

14:06:35 beach If I had had a bit more presence of mind, I would have discouraged the use of a call stack that automatically cleans up the activation record after a function call.

14:07:40 moon-child https://www.teamten.com/lawrence/writings/dynamic_cast_std_integer_c.html or so

14:07:42 jcowan I was married for 40+ years to a teacher, who was committed to the belief that everyone's mind can be changed. I was unable to share her view.

14:07:53 beach The CTO saw my point but didn't manage to convince the project leaders. Eventually the project was canceled with millions lost.

14:08:20 jcowan yes, unfortunately CTOs have authority but little power.

14:08:31 beach Indeed.

14:08:37 jcowan (of course, some have no authority either)

14:09:18 jcowan A friend of mine was a CTO for five years and described it as "a hideous trap"

14:09:29 beach Well, they are usually not trained in the domain, so they usually have to trust the project leaders and programmers. But this one at least was able to understand my argument.

14:13:45 jcowan hayley: That's easy to see: the Moon missions *worked*. Text rendering often fails.

14:48:06 younder My CTO at Opera was Håkon Lie - the creator of CSS. Maybe I was luckey

14:49:45 beach Are you saying CSS is considered a good thing?