freenode/#lisp - IRC Chatlog

8:14:04 holycow i have thought for a while now that someone could get a lisp os built and then use the linux kernel as an interface between it self and whatever hardware it wanted to talk to

8:14:19 holycow i'm talking out of my arse though, i have no idea about kernel programming

8:14:34 White_Flame I'd be perfectly happy with a fully user-space Lisp "OS" environment

8:14:46 shka yes

8:14:51 holycow well, mezzano is sort of on its way there

8:15:06 White_Flame given that you're not on highly constrained resources, the OS doesn't add a lot of practical overhead

8:15:11 White_Flame and gains you a lot of advantages

8:15:11 holycow proof of concept is kind of done .. well several times now i guess

8:15:18 holycow that was my thinking

8:15:29 holycow instead of writing the kernel in lisp, leave that to linux or whatever kernel team you want

8:15:34 shka mezzano does use separate kernel?

8:15:36 holycow and start consuming the userland

8:15:45 holycow shka: no, it does its own thing

8:15:53 shka rigth

8:15:59 holycow but in it's case it looks like a difference without a difference

8:16:19 holycow if you wrote your own userland in lisp, your 'kernel' would be interface between the userland and whatever kernel you used

8:16:38 holycow and if you used native linux virtualization, each users would be fully virtualized natively

8:17:21 holycow within the image you would still need to develop a security model so you would still need to have a kind of local kernel handling that

8:17:43 holycow you would want to give users the option to run a fully set of security contexts or none ... as they see fit.

8:18:21 attila_lendvai ACTION is very surprised by the Mezzano readme that claims to be an OS, but begins with a description of keyboard shortcuts...

8:18:30 shka let me wrap my mind around it

8:18:48 White_Flame why? its display manager is effectively part of the OS

8:18:53 holycow attila_lendvai: at the level of mezanno there is no difference between userland, kernel and the text editor really

8:19:42 shka holycow: i did not consider native virtualization, suddenly Lisp OS sounds a lot more reasonable

8:19:48 holycow only if the project grows will encapsulation be required

8:20:05 holycow shka: linux makes it totally doable. it is really an amazing kernel

8:21:02 holycow shka: i have sbcl running on this: https://nextthing.co/pages/pocketchip

8:21:08 shka so what we need is essentially kernel that has driver for linux

8:21:12 shka woooooooooaha

8:21:16 shka NICE!

8:21:53 shka anyway

8:21:59 holycow if i were to remove everything but the kernel and userland and throw mezanno on there as a virtual machine (someone far smarter than me would have to write that mezanno feature) you basically have that

8:22:37 holycow shka: that is how i see it. someone would write the linux kernel <--> sbcl interface and voila

8:22:38 shka you would not even need so much out of mezzano anyway

8:22:49 holycow i mean without using cffi ... maybe cffi is enough

8:23:22 holycow shka: mezanno is pefectly usable now. windows 3.1 doesn't work as well as mezanno and that had hundreds of paid programmers working on it

8:23:50 shka ok, this is probabbly by far easiest way to get Lisp OS for real world

8:24:08 shka damn, this can be so neat

8:24:14 holycow now we just gotta find some white knight with 100mil to throw around and pay some people to write this niche userland tha tno one will use

8:24:17 holycow heh

8:24:55 shka well, i think that lisp can be very well suited for building distributed system

8:25:05 shka you would have something that can work on almost anything

8:25:21 holycow well, there is a problem there

8:25:26 holycow if you think about the linux desktop

8:25:37 holycow gnome and kde are essentially what we are talking about

8:25:47 holycow they are a kind of runtime on top of linux kernel

8:26:18 holycow they simply assume they will be running on native hardware and have not really been "virtualized" ... and yet no one really uses linux out there

8:26:19 shka well, i was thinking more about remote controling of industrial stuff and things like that

8:26:52 holycow ah, right. that is probably the best case scenario as everything splinters. we end up in a niche of a niche somewhere

8:26:52 shrdlu68 I feel like something is brewing here. So much talk of Lisp OSes lately.

8:27:15 shka that is actually very important niche

8:27:39 shka relabile distributed stuff is really needed

8:27:44 holycow the iot stuff will probably take that over and create their own interfaces

8:27:50 holycow and do it poorly, as always

8:28:01 shka IoT stack is garbage

8:28:15 shka especially consumer IoT

8:28:18 holycow if they are cheap and and to the market first, they win, although i'm not a pessimist

8:28:23 holycow oh totally

8:30:10 shka anyway, my point is that it can be useful for many use cases

8:30:29 holycow possibly, i don't know enough to be honest

8:30:57 shka you could totally build cluster stuff as well but this crowd is all about java and python nowdays

8:31:07 holycow as long as we are thinking about deconstructing things, if you think about the pc, i was looking at my motherboard the other day and wondered why is everything soldered on to the mobo

8:31:19 holycow why is the sound chip, usb, cpu and all that on th emotherboard?

8:31:35 shka because why not?

8:31:39 holycow why isn't the mobo just a giant set of pcie16x ports

8:31:45 holycow and everthing is on an expansion riser

8:31:53 holycow yeah exactly, why not, it's always been that way

8:31:55 White_Flame cost

8:32:00 holycow and that right

8:32:05 shrdlu68 And why can't I hotplug a cpu like I can a gpu?

8:32:25 White_Flame you can hotplug your gpu?

8:32:29 holycow but then it occured to me, that i almost NEVER use the full power of my intel cpu. like ever, except gaming and some occasional 3d or graphics editing

8:32:33 holycow and there it is offloaded to the gpu

8:32:57 holycow what if the motherboard was just a set of expansion slots and your northbridge / southbridge was one card

8:33:18 White_Flame you'd probably get higher latency, higher power consumption, and lower speed as well

8:33:30 holycow and your processing cards were giant cards packed with arm cpus

8:33:33 holycow another for ram

8:33:40 holycow another for gfx and so forth

8:33:51 holycow hell the bootloader would be on a card

8:34:38 shka heh, i'm not so interested in hardware, really

8:35:24 shrdlu68 I like the idea. How practica is it?

8:35:59 holycow understood. just pointing out that we currently are living with some old designs because they work and as you guys said they are cheap. BUT! we also live at a time when a whole bunch of new things are already possible

8:36:10 holycow this is why people say things like above "i feel like something is brewing"

8:36:21 holycow amazing things are possible, we just don't have the money to play with any of it

8:37:14 holycow shrdlu68: no idea. but there is a 64 core arm based soc you can buy for 100 $ or so out there. i forgot the name

8:37:40 shrdlu68 Yeah I heard about it.

8:37:51 shrdlu68 I'll buy it one of these days.

8:38:00 holycow i cannot find the link but yeah, i think you know

8:38:21 White_Flame do you mean tilera or one of those?

8:38:21 holycow i have a pocket chip already, i pre-ordered one of these: https://pyra-handheld.com/boards/pages/pyra/

8:38:44 White_Flame the many-core streaming oriented chips aren't very general purpose

8:38:45 holycow here is the intersting part of the pyra, one sec:

8:38:59 holycow White_Flame: it had a different name, sec

8:39:11 White_Flame ah, the good ol' OpenPandora, which I never received

8:39:16 holycow here are the dev costs for pyra: https://pyra-handheld.com/boards/threads/money-makes-the-world-go-round.80207/

8:39:25 holycow that whole thing is built by one guy

8:39:36 holycow White_Flame: oh, you got hosed for your order? really?

8:39:45 White_Flame yep, many many years ago

8:39:52 holycow huh. did you get your money back?

8:39:55 White_Flame nope

8:40:00 holycow wow. why not?

8:40:26 White_Flame why would I? they ran out of money and pulled out of their obligations

8:40:35 holycow huh

8:40:38 holycow i did not hear about that

8:40:45 White_Flame tons of people got screwed over

8:41:05 holycow i will google that

8:41:21 White_Flame although some decided to pay hundreds of euros more to "upgrade" to buying one of the newer models when it transfered to the new dragonwhatever guy

8:41:25 holycow there are a ton of these kinds of things: https://www.pine64.org/?page_id=1194

8:41:48 holycow White_Flame: ohhhhhh ... really? huh.

8:42:37 White_Flame imagine a kickstarter failure except without kickstarter. same thing

8:42:38 holycow anyway, shrdlu68 i imagine it is doable but requires millions in dev costs to even contemplate.

8:42:50 holycow White_Flame: that is too bad.

8:43:11 White_Flame and the original guy ("greg"?) started playing ponzi schemes with orders before it collapsed

8:43:53 White_Flame ie, get new payments, use the money to get fewer older paid customers' item shipped

8:44:15 holycow if you look at the videos on youtube for the pyra, the design is really neat. the main module with keyboard is separate from the cpu. either could be upgraded independently. i can imagine the idea of 'modular' computing starting to actualy bubble up even though google gave up on the modular phone

8:44:33 holycow White_Flame: oh, aha. got in too deep and could not get out.

8:44:42 holycow sorry to hear about that.

8:44:45 White_Flame yeah, I think the design really hasn't changed since the openpandora days, except for parts that are no longer available

8:45:53 holycow White_Flame: yeah, so i'm looking to see how much of my desktop i can recreate on the pyra and the pocket chip

8:46:04 holycow i know i can run any linux login manager so we are good there

8:46:22 holycow my window manager is stumpwm and now that sbcl runs on arm i can run that on all of those devices

8:46:29 holycow the rest of the software seems to be there

8:46:40 holycow so they are all now keyboard driven interfaces so that should work

8:47:13 holycow i'm learning some more lisp so i can write a small display helper (chooser i guess) so it is easier to start apps on touch screens and small screens

8:47:47 holycow so kind of like that scene in terminator 2 where the chrome robot re-assembles it self, something is pulling it self together

8:49:14 holycow the biggest issue is lack of 3d acceleration, the graphics chipsets are proprietary so that is a bummer.

8:50:44 White_Flame so if I'm reading this correctly, the pyra still isn't released yet?

8:50:53 holycow correct, not yet

8:51:10 White_Flame wow, that's a loooong time

8:51:26 holycow it is. its one dude working on it. that dragon dude

8:52:07 holycow https://www.youtube.com/user/EvilDragon1717/videos <-- a few of the last videos show you the statuts of the device

8:52:14 White_Flame um, according to that costs page, it seems like he's farmed out a lot of work

8:52:59 holycow yeah. i've never built anything of that sort, i have no idea what is involved.

8:53:55 holycow i think the more he farms out the better. one person can only really do so much.

8:54:21 holycow apparently he found a company in the eu that specializes in keyboard materials so he could get that keyboard pad built

8:54:32 White_Flame but in any case, I think that phones/tablets have subsumed that fully portable form factor; keyboard is in theory nice, but I've worked on N900s and small bluetooth keyboards and such, and it's simply not really workable.

8:54:33 holycow i would have tried to do that my self and created some abomination

8:54:51 holycow White_Flame: i disagree for two reasons:

8:55:08 holycow a) i run blackberry devices. keyboard is a must, i can type faster than anyone on any keboard. seriously.

8:55:18 White_Flame um, no ;)

8:55:38 holycow :) i've not met anyone that can type faster, you may be the first

8:55:49 White_Flame you can do 100+ wpm on handhelds?

8:56:04 White_Flame that's not unreasonably fast

8:56:08 White_Flame for real keyboards

8:56:17 holycow no one can doo 100wpm on mobile

8:56:28 White_Flame right, so it's inferior in that form factor

8:56:31 holycow i can type faster than any ios or android user is all i claom

8:56:35 holycow claim even! hehe

8:56:55 White_Flame well, I'd say that's irrelevant, because "top of the poor usability pile" is still in the poor usability pile

8:57:03 joga I think I can sometimes do close to 100wpm on n900...

8:57:10 holycow b) i have one of these: http://www.ebay.com/itm/Rii-Mini-2-4GHz-i8-Wireless-Keyboard-Mouse-for-PC-XBox-360-PS4-Android-TV-Box-/151686600304

8:57:14 joga for short periods with simple irc babble

8:57:16 joga :)

8:57:21 holycow the one in that picture has a terible kb layout, i have one with a better faster layout

8:57:48 joga still waiting for replacement for it but using it for now since 8 years at least..

8:57:52 holycow White_Flame: not sure what the point is. for a mobile device you still need character input. you either do it with a physical kb or virtual

8:58:00 holycow phystical keyboard are actually amazing

8:58:09 holycow joga: that is impressive :)

8:58:19 joga using a virtual keyboard on a bus or something is pretty hopeless

8:58:27 White_Flame I've used tons of variations of touch keyboards and physical keyboards for many years

8:58:36 joga even on a big tablet it's unbearably slow and terrible

8:59:05 holycow joga: i agree. i workde hard to get virtual keyboards to work well. the only one that comes close to usable is on the iphone for me.

8:59:11 White_Flame the fact of the matter is that typing sucks in mobile/pocketable uses, so it's fine with the bare minimum anyway

8:59:20 holycow i just gave up and went to blackberry physical keyboards. there is no point in wasting time.

8:59:22 joga I use my phone 99.99% for ssh

9:00:13 holycow White_Flame: totally understand what you mean. however, that mini wireless kb is what i use at home on my pc sitting in my chair to ssh and do sysadmin things. i have one with fast key travel so its pretty good.

9:00:16 joga intarnet has become so bloated anyway the poor old maemo browser is struggling with even the simplest of pages because of all the javascript

9:00:27 holycow when i'm not able to type fast enough i switch to a regular keyboard

9:01:15 White_Flame holycow: yeah, I've been looking for a nice floating keyboard/mouse combo for the living room, where occasional use is fine for the RPi. However, that's not mobile use, where you're schlepping around both your screen-based device and external keyboard like that

9:01:17 holycow so it is doable, the keyboard on mobile has not gone away imho. it works great. for users that just do simple texting android virtual kb is fine or whatever.

9:01:47 holycow White_Flame: well no, i think we are miscommunicating. we are talkinga bout different things.

9:01:59 joga something like this might be usable for my purposes though it's "bulky" https://www.indiegogo.com/projects/gpd-pocket-7-0-umpc-laptop-ubuntu-or-win-10-os-laptop--2#/

9:02:03 holycow my blackberry has a keyboard. so its a physical keyboard on a mobile device

9:02:35 White_Flame right, I was referring to stuff like the wireless keyboard you linked

9:02:35 holycow the pyra has 3g/4g built in ... so the moment that becomes viable, the pyra will take over my sysadmin stuff from a mobile device.

9:02:53 holycow the phoen will just become for phone calls, i will no longer care about if my phone has a keyboard.

9:03:03 holycow White_Flame: oh no! hehe!

9:03:17 joga this was also interesting but seems the people behind it may not deliver and linux support is questionable... https://www.indiegogo.com/projects/gemini-pda-android-linux-keyboard-mobile-device-phone#/

9:03:21 holycow no that is useless for that, that is for home. i'm too lazy to use a proper keyboard when sitting on the couch.

9:03:32 joga but at least they recognize the need for such a device

9:03:32 holycow that mobile keyboard controls the tv, computer, everything for me.

9:03:52 holycow joga: looking

9:04:27 holycow joga: yeah its great they are doing that BUT ... i can tell you why that will fail

9:04:33 holycow the keyboard is NOT thumb oriented

9:04:36 holycow it is a mini kb

9:04:39 White_Flame which is a good thing

9:04:39 holycow that is going to be hard to use

9:04:46 holycow i know, i had something like that 10 years ago

9:04:50 White_Flame I had a Jornada 720 back in the day, probably the best keyboard on such a device I've used

9:04:52 holycow i still have the device

9:05:08 holycow with that exact size of keybaord keys

9:05:39 holycow is separate keyboard use into two distinct categories: thumb oriented + palm oriented

9:05:52 holycow both are compeletely different types of use cases imho

9:06:11 holycow if you cant use the gemeni with a thumb you are not going to really get much out of it

9:06:20 White_Flame thumb is dumb ;)

9:06:33 holycow heh it works fine for certain things.

9:07:10 holycow the gemeni looks awesome tho, thx for the linkage joga

9:09:38 holycow joga: i cannot find the link, but i have an old asus laptop almost exactly the same size as the gemeni

9:09:58 holycow it was hard to use because it was just big enough that you could not thumb it, just small enough that typing was super cramped.

9:10:00 argoneus hello

9:10:24 argoneus can I discuss lisp IDEs here?

9:10:27 argoneus or just the language

9:10:53 beach argoneus: You can discuss IDEs.

9:10:56 White_Flame this channel is Common Lisp oriented, so if you're talking about Scheme or something it's better off in #scheme, etc

9:11:05 White_Flame *Scheme IDE

9:11:07 joga holycow, check this also for cons... https://talk.maemo.org/showthread.php?t=99022

9:11:10 argoneus I was wondering which Common Lisp IDE was recommended for a beginner on windows

9:11:17 joga ACTION stops off-topic conv :)

9:11:17 argoneus I found multiple, lispworks looking the best, but it's paid

9:11:25 White_Flame Emacs is the best supported environment for Windows

9:11:34 argoneus oh, huh

9:11:35 White_Flame it's ... different, but it's worth learning for the Lisp support

9:11:45 argoneus oki

9:11:49 White_Flame actually, remove "for Windows" from the previous statement :-P

9:11:57 argoneus can I use the same environment for other dialects too?

9:12:01 argoneus like if I decided to learn clojure or such

9:12:15 argoneus or is it specific to common lisp

9:12:21 White_Flame I think Clojure might have good support under Eclipse, not sure

9:12:33 argoneus alright, I'll give it a shot, thank you!

9:12:49 White_Flame in any case, it's a good thing to be able to hop between environments and do the idiomatic thing that the language expects, instead of trying to lock into 1 thing and doing everything through that

9:13:21 argoneus "SLIME" is what I want, right

9:13:25 White_Flame if you want to get started with Common Lisp, you should have Emacs, SLIME (which interfaces between Emacs and the running Lisp), and Quicklisp to download libraries

9:13:33 White_Flame yep

9:13:41 argoneus thanks!

9:19:52 holycow White_Flame: the thing about the thumb keyboards is this: no two people can agree on the optimal keyboard layout

9:20:15 holycow once you shrink a keyboard down, you split usability down to a single usecase per person.

9:20:30 holycow that is probably the real problem with thumb keyboards

9:21:21 White_Flame that's an argument for screen keyboards, that they can be reconfigured

9:21:32 holycow that is the thing i'm really looking forward to with the pyra. theoretically, one ought to be able to get thumb keyboards built for their own needs ... although still stupidly expensive to do

9:21:43 holycow White_Flame: i agree with that, absolutely

9:25:14 holycow joga: great read

9:29:28 argoneus hm

9:29:51 argoneus when I write sbcl, it only prints that it's free software but there's no interactive console

9:29:55 argoneus to actually install quicklisp

9:30:39 White_Flame ideally, you'd be launching sbcl from inside emacs

9:30:54 White_Flame but you're at an interactive prompt where you can run the quicklisp installation stuff anyway

9:31:07 argoneus thing is, it's not interactive

9:31:09 White_Flame the "* " is the default prompt here. Type (+ 1 1)

9:31:12 argoneus no matter what I write, nothing happens

9:31:14 argoneus there's no *

9:31:22 argoneus just the license stuff

9:31:35 White_Flame maybe you typed some weird control characters first? kill it and try again

9:31:42 argoneus been trying for 15 minutes

9:31:49 White_Flame what platform are you on?

9:31:54 argoneus ubuntu

9:32:01 argoneus decided to use the ubuntu thing in win10

9:32:02 White_Flame how did you install sbcl?

9:32:06 argoneus apt-get install sbcl

9:32:11 White_Flame ah, I have no idea how its linux-in-windows stuff works

9:32:21 White_Flame get ye to emacs

9:32:22 argoneus it's not emulation or such

9:32:26 argoneus it's a proper ubuntu

9:32:39 argoneus can run a webserver on it if I decide to

9:32:45 argoneus I'll try through emacs (never used emacs before)

9:34:37 White_Flame emacs+SLIME in particular

9:34:51 argoneus don't I need to install quicklisp first to get SLIME working?

9:35:23 White_Flame I honestly don't know which is the best to install nowadays

9:35:29 holycow thanks for the inferno heads up earlier. this is a great watch: https://www.youtube.com/watch?v=3d1SHOCCDn0

9:35:32 White_Flame rem, which order to go

9:36:05 White_Flame but it could be that if sbcl doesn't even start up properly, that it won't start under SLIME either

9:36:30 White_Flame my suspicion is that the terminal IO is funky and causes it not to work; shelling to it directly from emacs might alleviate that

9:37:36 White_Flame you can do sbcl --load quicklisp.lisp --eval '(quicklisp-quickstart:install)' and see if it will run scripted like that

9:37:41 White_Flame for the bootstrap of quickly

9:37:49 White_Flame *quicklisp

9:40:27 argoneus that just prints the copyright info again

9:40:27 argoneus hmm

9:41:27 holycow argoneus: i cannot help you on this. i have no idea about windows 10, especially the ubuntu thing they have

9:41:46 argoneus it's an actual regular ubuntu inside it

9:41:56 White_Flame at least, that's the marketing spiel :-P

9:42:11 argoneus I can run llvm and all I want on it easily

9:42:11 argoneus :(

9:42:22 argoneus llvm is normally a pain on windows

9:42:25 holycow k. so what is the problem then?

9:42:31 argoneus the repl just, doesn't work

9:42:46 White_Flame I'd say rebuild sbcl from source locally and see if it's any different

9:42:48 argoneus thing is, if I press ctrl+c in the REPL, I get a debug thingy

9:42:53 holycow apt-get install sbcl emacs24-nox slime should give you everything

9:43:36 holycow you want to use emacs nox and work out of a terminal instead of emacsx. emacsx seems to frustrate users of normal graphical interfaces

9:44:06 holycow once you have those 3 packages, run emacs then m-x slime and you will get a connection to sbcl

9:44:10 White_Flame (graphical emacs should be more responsive. on a local box...)

9:44:43 holycow then in emacs you can do c-x 1 or c-x 2 or c-x 3 to split up your shell window so you can code in emacs and see the slime at the same time

9:44:53 holycow c-x 0 collapses all emacs windows

9:44:58 holycow emacs frames i mean

9:45:08 holycow that is your short 10 second intro to get it up and running argoneus

9:46:28 argoneus the repl is just dead even inside emacs

9:46:41 White_Flame rebuild sbcl from source

9:46:44 argoneus unless I'm missing something

9:46:54 holycow what do you mean dead?

9:46:58 holycow post a screenshot online

9:47:03 argoneus http://i.imgur.com/71cI5FJ.png

9:48:12 holycow aha

9:48:17 holycow its stuck at the compile part

9:49:02 argoneus try building from source, then..?

9:49:14 holycow argoneus: my apologies i am not going to be of any use beyond this. i would just run ubuntu in virtual box instead

9:49:18 holycow i know that it works on there for sure

9:49:28 holycow i bet ubuntu in windows 10 is not bone stock at all.

9:50:00 holycow i would say it is quicker to download ubuntu iso and run it in a virtual box vm than building from scratch

9:50:09 argoneus hmm

9:50:18 holycow building from scratch means you are still in a microsoft bastardized version of ubuntu

9:50:20 White_Flame my laptop died, so I'm faced again with having to pay the microsoft tax on hardware before reinstalling linux over it. I hate that fact.

9:50:25 holycow there is no way microsoft did not fuck with that

9:50:27 White_Flame but nobody should be using Win10

9:50:34 White_Flame especially as a developer or power user

9:51:01 holycow White_Flame: easy fix. only buy used hardware. someone else paid the tax and you get a know working unit, no new components to worry about :)

9:51:26 White_Flame I need some oomph, though, so not looking at older models

9:51:30 holycow argoneus: yeah, i think you are trusting microsoft too much. get a plain ubuntu image run it in a vm and ssh with putty into it.

9:51:34 holycow it works fine, i've done it lots

9:51:45 argoneus for some reason whenever I try to ctrl+c out of the prompt

9:51:49 argoneus I get a debugger thing

9:51:53 argoneus that seems to work and run commands

9:51:56 argoneus hm

9:52:16 White_Flame that's normal

9:52:21 White_Flame use ^D if you want to exit

9:52:46 argoneus "debugger invoked on a TYPE-ERROR in thread..."

9:52:47 argoneus eh

9:53:23 White_Flame the thing is, environments like CL are generating their own machine code on the fly, and run their own ABI internally

9:53:34 White_Flame I could very easily see the lin-on-win stuff meddling in that heavily

9:53:58 White_Flame especially when it comes to signals, terminal IO, memory management, etc

9:54:19 argoneus mm

9:55:15 argoneus ha ha ha

9:55:21 argoneus I downloaded the sbcl binaries from the website

9:55:24 argoneus and those work just fine

9:55:29 argoneus I don't understand anything anymore

9:55:57 argoneus seems like the ubuntu repo version is somehow borked

9:56:46 argoneus though I guess now I need to install quicklisp and slime manually

9:57:39 White_Flame get quicklisp to install slime, via quicklis-slime-helper or somesuch, should be on the quicklisp page

9:57:53 White_Flame from there, you can launch from emacs

9:58:09 White_Flame (assuming that all the paths aren't confused from being inside windows)

9:59:44 holycow argoneus: i just download ubuntu desktop 14.04, installed the vm, installed emacs23-nox sbcl and slime and got slime up and running

10:00:05 holycow you are wasting your time with that microsoft nonsense.

10:00:18 argoneus I think it's some bug in ubuntu itself

10:00:22 argoneus or rather the packages in the repositories

10:00:24 holycow argoneus:

10:00:35 argoneus I don't see a reason why a binary from apt should be different from the one I downloaded

10:00:44 holycow i just download 14.04, installed it in virtual box and got emacs and sbcl running in less time than you complaining here

10:00:49 argoneus ;;

10:00:51 holycow please get a fucking clue and stop wasting your and our time

10:00:58 argoneus b-but I just got it working

10:01:04 argoneus just trying to understand why

10:03:13 holycow because one is probably compiled to be multithreaded and the other singlethreaded

10:05:36 argoneus http://i.imgur.com/PMr7izX.png this is correct, right

10:05:56 holycow yup

10:06:00 argoneus thank god

10:06:02 argoneus thanks for the help!

10:10:02 argoneus now to just figure out how emacs works, heh

10:15:07 argoneus oh, emacs seems a lot more intuitive than vim

10:15:35 holycow welcome to the dark side

10:17:23 holycow thanks to all for the mention of k-machines

10:17:29 holycow i've never heard of the project

10:17:34 holycow here is a great articlet on it: http://fare.tunes.org/tmp/emergent/kmachine.htm

10:18:08 holycow apparently the author managed to get a hold of the k-machine hardware after the ownership collapsed into bankruptsy

10:20:19 shka sooo

10:21:03 shka i kinda want to figure out how to develop kernel image that can be loaded directly with KVM

10:27:44 p_l holycow: the author is one of the designers of K-machine

10:28:00 flip214 shka: your best bet might be "-kernel bzImage", else you'll need a bootloader too

10:28:14 shka yes

10:28:21 p_l so long as early bootstrap is compatible with bzimage format it will work

10:28:22 shka i am knew at that

10:28:27 shka *new

10:28:43 shka never even considered doing something like that before

10:28:55 p_l you might need to write some assembly that gives you the bootstrap necessary to start whatever lisp image you need

10:29:10 holycow p_l: neat!

10:29:25 shka oh, god, please just no x86 assembly

10:29:32 shka -_-'

10:30:05 shka anyway, i need to first figure out how to write the simplest kernel

10:30:43 shka but anyway, this can be really usefull tool

10:31:34 shka if only just for ultra light sandbox deployment

10:31:56 argoneus aaa this buffer thing in emacs is great

10:32:06 argoneus thanks so much for helping me earlier

10:32:24 p_l shka: generate the assembly from Lisp

10:32:39 p_l or you might find out that loading raw kernels in KVM is maybe not the best idea ;)

10:33:02 shka can you be more informative?

10:33:20 shka i have no idea what i am doing right now so any hints are welcome

10:33:32 p_l loading bare kernel you lack all kinds of services that might be provided by the environment you start in

10:33:54 shka yes

10:34:02 shka that's obvious

10:34:06 flip214 shka: TBH, I'd go for a (docker-like) jail (using chroot and namespace support, virtual networking, etc.) and an lisp running therein as sandbox...

10:34:31 flip214 else you'll need to reimplement device drivers, filesystems, memory allocation (no mmap for you!), etc.

10:34:42 p_l with let's say UEFI system you instead get a boot environment programmable in high level language (pretty much any language can be used) which can limit the need to interface with assembly even less

10:34:51 flip214 but if you really want to go that route, there's Mezzano

10:34:52 shka can't i just use host stuff?

10:35:00 flip214 which host?

10:35:02 shka iirc that's the point of KVM

10:35:05 flip214 the kernel has no host.

10:35:08 p_l shka: you still need drivers

10:35:14 shka i don't want to write bare metal kernel

10:35:17 flip214 yeah, you can write drivers for the virtual hardware...

10:35:35 shka just something that sits on top of KVM

10:35:42 p_l KVM emulates a bare metal environment with possibly some esoteric virtual hardware (like virtio-pci devices)

10:36:27 p_l it would be much easier to make a driver for virtio 9fs than a full block device stack + filesystem, though

10:36:42 shka i see

10:37:10 shka can we talk about that later?

10:37:23 shka i think we are getting somewhere

10:40:35 holycow p_l: did you just suggest that shka write a driver for 9fs over virtio?

10:40:52 p_l simpler than many other options

10:41:00 holycow if i'm getting what you are suggesting that is a diabolically genious suggestion

10:41:01 holycow jesus

10:41:25 shka *buahahahaha*

10:41:34 holycow no seriously, i never thought of that, that is amazing

10:41:57 p_l virtio bus is simplified, and for 9fs you only need essentially a simple message passing buffers

10:41:58 shka anyway, this is really exciting but i have bug at office and boss is not amused ;-)

10:42:01 shka back to work

10:42:16 holycow p_l: i emailed joe marshall, i'm curious to learn more about that project. that is so fascinating. thank you for the tip.

10:43:23 holycow this channel blows my mind. unfortunately it is always on bloody european time. i'm not supposed to be awake right now.

10:48:40 p_l heh, I opened the old CONS memo and found similar arguments for "simple unspecialized architecture" as I made :D

10:49:52 p_l "Using a very unspecialized processor was found to be a Good idea for several reasons."

10:50:01 holycow heh

10:50:02 p_l ym: ^

10:50:16 holycow broke your mojo there, sorry

10:50:27 p_l AI MEMO 448, LispM group report from MIT

10:50:33 holycow just reading joe marshalls blog. very very smart dude.

10:54:24 holycow p_l: it totally makes sense. we are always slaves to the resources available in our curent time.

10:54:51 holycow in the 70's and 80's specialized cpus seemed like the way forward as generalized processing was not established as it is today

10:55:04 p_l you'd be surprised

10:55:14 holycow tell me more

10:55:33 p_l it's more that there was a *lot* of ISAs, and a lot of assembly-level programming (and less proliferation or standardization between them)

10:56:05 holycow oh, so nothing has changed then

10:56:07 holycow heh

10:56:10 p_l also, I guess some of it was just like the use of extra dedicated languages, something often frowned about today but which was common then

10:56:37 p_l so using a bunch of the same chip in different parts of the system running specialized software was not that uncommon

10:56:57 p_l I think some of the IBM peripheral processors were essentially the previous generation mainframes running dedicated software

10:57:36 p_l today, each IBM z "processor book" has one more CPU which runs special microcode instead of normal software and essentially implements some of the "hardware" that everything else uses

10:59:24 p_l when main cpus weren't powerful enough, a lot was done through dedicated hw, then main cpus got faster, then some dedicated hw leapfrogged them, and so on and so on

10:59:32 holycow right right

10:59:56 p_l Take for example audio - today, most people have audio hw that is less powerful in terms of computation than what I had in my first PC

11:00:14 holycow yeah its all in the driver now

11:00:16 shka yes

11:00:27 shka almost no hardware mixing

11:01:00 p_l holycow: because in the past, having a 16 channel hw mixer was necessary to actually have 16 sounds mixed together - or for example to have music through MIDI

11:01:28 shka p_l: what is 9fs?

11:01:35 shka and what is virtio?

11:01:41 holycow p_l: i watch the 8bit guy on youtube. he actually explaine that (but for like 8 channels i think) on some popular chip in 80's consumer hardware

11:01:42 p_l and the systems that didn't have dedicated processing hw but had present-day style audio, tended to have it single-user only

11:02:35 p_l shka: virtio is a set of buses and devices, implemented among other things in KVM (by linux kernel) or QEMU (for emulated devices) and others. They are simplified because they don't have all the trappings of physical device

11:02:47 p_l so you have much easier time writing drivers

11:02:58 shka i see

11:03:04 shka what about 9fs?

11:03:08 p_l 9fs, or rather 9P, is the Plan9/Inferno filesystem, based on messaging

11:03:16 shka i see

11:03:22 p_l well, RPC, sort of

11:03:23 holycow read about plan9 and watch youtube videos

11:03:26 holycow it will blow your mind

11:03:32 shka CSP style?

11:03:45 p_l it's a network filesystem

11:04:01 p_l but it's messages map 1:1 to Plan9 i/o syscalls

11:04:15 shka ok, that's enough for now

11:04:26 p_l anyway, there exists a specification for transporting a 9P connection over virtual PCI device

11:04:38 holycow the other crazy thing you might want to consider is looking at porting sbcl to plan9 / inferno

11:04:58 p_l and QEMU provides a way to export a directory as filesystem accessible over that virtual PCI device

11:05:00 holycow and host that in a kvm vm

11:05:16 p_l so instead of writing a FS, block drivers, etc, you can use files directly :)

11:05:36 shka that's what i want

11:05:48 shka if it is networked, it is even better

11:08:18 shka holycow: actually i am under impression that in ideal world i would get plan9, smalltalk enviorment, make it one and lisp and be blown away by the awesomness

11:09:02 holycow shka: if we are allowed to imagine such things without people telling us why it is a bad idea, it sure is sounding intersting.

11:09:04 p_l technically you could export Plan9's network API over 9P to provide quick'n'dirty network support till you write a network stack in lisp

11:09:22 p_l (an old trick for Plan9)

11:09:27 holycow plan9 theoretically doesn't care about the underlying hardware, its purely a namespace + utilities

11:09:55 holycow if you can let linux care about hardware ... hmmm. i have never thought about this before

11:10:03 shka neither did i

11:10:37 holycow p_l: what have you started

11:10:52 holycow i mean, LOOK at what you have started

11:10:53 holycow heh

11:10:57 p_l holycow: If I wanted Linux to take care of hardware completely, I'd go even weirder way ;)

11:11:10 holycow okay, blow us away once more

11:11:14 holycow i've got the popcorn

11:11:29 shka https://www.youtube.com/watch?v=pitwR8jPd_M

11:11:30 holycow and a lisp programming video explaining object oriented features

11:11:30 holycow heh

11:11:34 phoe where the hell am I

11:11:37 phoe I just logged back on

11:11:57 holycow phoe: i have no clue my self. this channel is amazing.

11:11:58 shka phoe: hey, we are just building lisp os on top of linux thanks to plan9

11:11:59 phoe and landed in #plan9 or #alternative-os

11:12:10 shka full hipster mode engaged

11:12:35 p_l L4/Linux, with some kind of LispOs working as one process, and linux kernel running as another process, with servers providing use of the linux drivers to Lisp process over L4 IPC calls - which means that as your system matures, it can bring over more and more code inside

11:12:47 holycow huh

11:12:49 holycow wow

11:12:58 p_l including possibly implementing the L4 API in lisp at one point and replacing the previous L4 implementation with your Lisp-based one

11:13:02 holycow p_l: well then, how about we mate that with my pc architecture idea:

11:13:34 holycow * the motherboard is just a set of pice16 slots (or something like that). first slot contains a card with the bios, north bridge, southbridge

11:13:52 holycow all the other cards are either full of arm chips, sound cards or whatever

11:13:59 holycow and throw your idea on that and holy crap.

11:14:35 p_l holycow: wouldn't work, as PCI-E is not a backplane parallel bus

11:14:47 holycow okay, i knew there was a problem

11:14:56 holycow your approach is much more elegant though, i like it.

11:15:00 p_l holycow: what you described is however how one would build a computer back in time of ISA, UNIBUS, VME etc.

11:15:13 p_l with passive backplanes

11:15:31 holycow but, of course it was already done. why would anyone actually think this wasn't done 40 years ago.

11:15:38 holycow heh

11:15:47 p_l I think there was even a frankenstein monster of two TI Explorer machines stuck in one card cage

11:15:59 holycow huh

11:16:34 p_l holycow: by carefully specifying who had access to what addresses on the bus, they could share one card cage and even communicate over shared memory

11:16:42 holycow huh

11:17:31 p_l (DEC did similar thing in software with their partitioning support in Alpha, which could run multiple VMS and Digital Unix systems simultaneously by simply not telling them of hardware that was allocated to others)

11:17:32 holycow i have been looking at our servers for years now and thinking that instead of inserting hard drive trays into the backplanes, could that be generalized so that you can input cards for various functionality.

11:17:44 holycow obviously, no one needs this otherwise it would exist.

11:17:54 p_l this of course works only if the software is well behaved

11:18:14 p_l holycow: passive backplane systems are lousy in terms of performance, which is why we no longer use them even for disks

11:18:26 holycow but then, you could have backplanes on either side of the case. one the back you insert backplanes for whatever connector type you need (hide the cables) on the front you can have cards that add am/fm radio and whatever.

11:18:33 p_l (instead using point-to-point links, even if located on a "backplane")

11:18:33 holycow aha.

11:18:58 holycow *nod* of course this was all done before. it is so frickin weird.

11:19:00 p_l SATA, SAS, PCI-E are all network buses (SATA sucks, though)

11:19:05 holycow IT even

11:19:23 p_l SAS, PCI-E (and DisplayPort, too) are all based on network passing messages, with routers etc.

11:19:29 holycow oh!

11:19:41 holycow no kidding? i had no idea.

11:20:05 p_l holycow: at low enough level, all three of them sent around small packets of data with addresses ;)

11:20:33 p_l (PCI-E and DisplayPort even share the same physical layer stuff, differing only in higher level protocols)

11:20:40 holycow it makes sense the way you type it, i had no idea the packets had an address space. i thought it was all direct.

11:20:54 holycow but if you think about it, of course it can't be direct. there is too much going on.

11:21:22 p_l depends on how you build it, but without addresses, you can't have switches, so you'd be limited by amount of links a single chip can have

11:21:35 holycow right. that makes sense.

11:23:38 p_l holycow: most modern CPUs are also networked between themselves to support multi-cpu, instead of using common shared bus

11:23:47 holycow that makes sense

11:24:08 holycow a couple of years ago, a programmer blew my mind when he explain that x86 is no longer x86 underneath either

11:24:11 p_l holycow: on PCs, this stuff arrived with AMD K8, then showed up with Nehalem for intel

11:24:25 holycow x86 is just the api to a different architecture underneath

11:25:04 p_l holycow: well, depends on how you describe it, but in general all x86 cpus (I think ever) were microprogrammed even if it was 1:1 from x86 instructions to internal component states

11:26:01 p_l holycow: most of the "big" CPUs in the older times were microprogrammed, some quite extensively

11:26:13 fiveop `sbcl --non-interactive --eval "(dotimes (i 2) (print i))" | head -n 1` results in a broken pipe error in sbcl. That makes sense to me. How would I go about getting a common lisp (or probably any) application to handle that gracefully?

11:28:15 fiveop Having asked the question, I get an idea. A HANDLER-CASE for STREAM-ERROR might be a good starting point.

11:28:25 p_l holycow: that was often because they were made from much different components - for example, KS10 cpu (on which a lot of lisping was done) was made from 4-bit chips from AMD (for 36bit system)

11:28:52 holycow thats so weird

11:29:07 holycow maybe not trully weird, it just sounds odd today

11:29:27 flip214 well, every ALU starts with half-adders....

11:31:27 p_l holycow: you could chain ALUs to provide longer word, and you'd have a control store and some circuitry that would take an instruction from main memory, locate its definition in microcode memory, and execute said microcode (often by simply pushing the microcode memory outputs onto "operation" inputs of the ALUs etc.

11:31:52 holycow yeah, i just realized how stupid my statement is

11:35:51 holycow okay so, we have p_l's super weird os. step #2, get $100 mil and fund the development

11:36:16 holycow bill gates is wasting his time trying to save billions of people from tuberculosis. he should be funding this.

11:36:19 holycow heh!

11:36:41 holycow p_l: thank you for the great tutorial. i owe you a beer or two.

11:43:49 holycow and figured out how to save this conversation. thanks all.

11:48:24 shka beach: do you have any video explaining step by step what is SICL?

12:03:12 beach shka: Sorry, I don't.

12:03:24 beach shka: But I'll be happy to explain it to you here if you want.

12:03:42 shka i think i will just read docs back at home

12:03:48 phoe beach: Need any help making one?

12:04:03 shka but thanks for offer :-)

12:04:10 beach phoe: I think the demand is very limited at the moment. But thanks for the help.

12:04:13 phoe Since I think that SICL is advanced enough for such a video/presentation to be created.

12:04:18 phoe Okay.

12:04:42 phoe So, since shka passed on the offer, I'll gladly accept the explanation from you, beach :)

12:05:02 shka beach: we had discussion about smalltalk style version control software in lisp, and SICL just poped up

12:05:08 shka because of envs

12:05:19 p_l and first class environments

12:05:26 shka yeah

12:05:35 beach phoe: Here and now?

12:05:36 shka at #lisp-pl that is

12:05:48 shka (join us, we have cookies)

12:07:09 presiden (sponsored comment)

12:07:44 shka sadly not

12:08:01 shka i am not getting any money from promoting lisp

12:08:07 Josh_2 Did someone say cookies?

12:08:18 phoe beach: here and now, hm.

12:08:21 phoe Depends.

12:08:28 phoe Would you do it through IRC?

12:08:36 beach I could try.

12:08:43 beach But we might bore people to tears.

12:08:45 phoe Or would we need to get some kind of audio (and video?) chat?

12:08:58 phoe beach: haha, well, we might do it in query in this case.

12:09:10 beach What kind of query?

12:09:17 phoe /query beach

12:12:34 beach phoe: Basically, I think of SICL as a collection of modules, implementing various parts of the Common Lisp standard. The difference between those modules and existing modules in other implementations has to do with modern programming practice, test, documentation, and especially implementation-independence.

12:13:44 beach phoe: Now, some people have interpreted this collection of "modules" to mean that they can write a primitive Common Lisp in (say) C, and then start loading SICL modules to obtain a complete implementation. That is not the case.

12:13:51 holycow are there any changes to the spec at all? or is it just a re-organization with addong tools?

12:14:00 holycow addon even

12:14:37 beach phoe: The `modern programming practice' part involves using the full Second Climacs language, and especially more generic functions that most equivalent existing modules do.

12:14:44 beach holycow: For me?

12:14:57 beach holycow: No changes to the spec, no.

12:15:06 holycow yes sir. may apologies for interrupting, i did not realize you are in a stream of thought

12:15:14 beach No, it's fine.

12:15:20 beach Interrupt as much as you like.

12:15:33 phoe holycow: it's IRC, we interrupt each other all the time and most of us don't really care and just keep typing

12:15:48 beach phoe: A large part of SICL is Cleavir, an implementation-independent framework for writing Common Lisp compilers.

12:16:02 phoe beach: what do you mean by the "full Second Climacs language"?

12:16:08 beach OOPS

12:16:17 beach the full Common Lisp language

12:16:22 beach my abbrevs. :)

12:16:52 p_l beach: one kinda needs full CL first to load the stuff and probably use Cleavir to write a code-generating compiler, right?

12:16:53 phoe Got it. So, standard Common Lisp with a lot of pressure on generic functions, correct?

12:16:59 beach phoe: Yes.

12:17:23 holycow someone please log this, then update the wiki for the git repo for SICL pls

12:17:25 beach phoe: I don't think it has ever been attempted before to write an implementation-independent compiler for Common Lisp.

12:17:35 phoe holycow: #lisp is automatically logged.

12:17:46 holycow oh! perfect.

12:18:04 phoe holycow: http://ccl.clozure.com/irc-logs/lisp/

12:18:15 beach p_l: Yes, I am working on the bootstrapping part. I am using SBCL for that.

12:18:50 beach So, one example that appears to baffle people is that SICL LOOP uses LOOP for its implementation.

12:19:34 beach But there is no contradiction here, because the LOOP macroexpander runs at, well, macroexpansion time, which will happen in the host Common Lisp system during bootstrapping.

12:19:46 beach After that, there is no trace of the LOOP macro.

12:21:01 p_l beach: I proposed, in discussion with shka and phoe, that SICL could be used as base to implement extensions to existing CL implementations, maybe providing impetus for CDR process by making it easier to implement proposals

12:21:03 phoe beach: haha, I asked that question to you at ELS16. I'm still amazed by the explanation I got from you back then.

12:21:07 beach One of the first modules I started was the "sequence functions" module. It has taken me three attempts to finally have an acceptable strategy for how to do it. And I haven't even finished implementing the module according to that strategy (which is in the ELS paper this year).

12:21:40 beach p_l: That sounds possible, yes.

12:22:12 p_l huh, I see much more CDRs published than I have seen last time

12:22:39 beach My secret hope is that the SICL modules will be better than the native ones in many respects, so that implementations would prefer the SICL ones. The net result would be that we can cut down on total maintenance of all Common Lisp implementations.

12:23:02 beach p_l: "many". Many things, much stuff.

12:23:09 holycow what an interesting idea

12:23:16 p_l heh, right

12:24:10 p_l beach: I wonder how amenable to community extensions the commercial lisps would be. While some consider it "not a problem", I would prefer to not have a split between commercial lisps with their proprietary libraries and open source world

12:24:11 beach p_l: I think that's the essence of the idea. The only remaining part is that, once I have all the modules (I pretty much do), there would be no reason to refrain from creating a complete Common Lisp implementation out of them.

12:24:35 beach Er, make that phoe.

12:24:52 p_l beach: arguably it could be even a hosted implementation, kinda like "future-common-lisp" on Symbolics etc.

12:25:44 beach p_l: I haven't really given any thought to extensions. I have given some thought to an updated standard, but it would only fix obvious bugs and especially define many things where the current standard leaves them undefined.

12:26:51 beach phoe: Does that give you the information you wanted?

12:27:23 p_l I was thinking mostly of extensions that would "open up" hackability of the implementation - like user-extensible hashtables from CDR#2, extensible sequences, etc.

12:27:38 beach Yes, I understand.

12:27:53 phoe beach: yes, that's a good explanation.

12:28:38 beach p_l: One thing might interest you then. I think in SICL, because I have faster generic dispatch, fewer things would be made out of ordinary functions and built-in classes, and more things out of generic functions and standard classes.

12:29:25 beach p_l: That way, it would be easier to create subclasses, or entirely different classes participating in the same protocol.

12:30:15 p_l I was thinking of something like that for other implementations, possibly with some extra-fast-path for normal calls and extension mechanism for others

12:30:47 beach That might be necessary in implementation that insist on keeping the slow, native, generic dispatch. :)

12:33:57 shka beach: so basicly, i need full lisp implementation (let it be ecl since it seems to be very portable) to bootstrap actual implementation, right?

12:34:42 loke___ shka: Which actual implementation do you want to ootstrap? SBCL?

12:34:55 beach shka: Yes. Though right now, the bootstrapping is not complete, so there is no point in trying.

12:35:07 shka some kind of SICL based implementation ofc

12:35:17 loke___ I've always turned to ABCL when boostrapping SBCL.

12:35:18 beach loke___: You came in late. We are talking about SICL.

12:35:27 loke___ Oh. sorry.

12:36:00 shka so SICL will implements language syntax, semantics and even most of the compiler

12:36:11 shka as modules

12:36:54 shka ok

12:36:55 beach shka: That's the plan. The compiler is a big, customizable module though. The word "framework" is more appropriate. And it exists. Clasp is using it for its good compiler.

12:37:08 shka right

12:37:10 shka great

12:37:28 shka now i understand what you are trying to do

12:37:40 shka that is very good plan

12:37:45 beach Thanks!

12:38:22 shka and even more

12:39:12 shka once this will be adopted (and it should get adopted) we will be able to actually extend language by providing additional modules

12:39:12 beach Cleavir, the compiler framework, is particularly interesting I think. It uses CLOS to allow implementations to customize it for their needs. And I plan to implement a bunch of standard compiler-optimization techniques on intermediate code, plus some of my own.

12:39:16 phoe loke___: the logs are available, you can read up

12:39:29 shka and implementations will be free to either include those or not

12:40:03 beach shka: That should be possible, yes.

12:40:10 holycow this is so meta

12:40:32 shka well, we need some central point

12:40:46 phoe beach: holycow: http://cliki.net/SICL

12:40:51 shka we got one, it is called quicklisp and it had awesome effect on lisp popularity

12:41:14 phoe shka: we will have CLUS that I expect to be another central point when it comes to docs.

12:41:32 shka yeah, this is also needed

12:41:34 phoe beach: I allowed myself to become an editor and edit the IRC logs into a description of SICL.

12:41:54 beach phoe: That would be fine with me.

12:42:11 phoe beach: Good - then the CLiki SICL page is up and running.

12:42:20 shka and once we will have SICL running everywhere we will be in very healthy place

12:42:44 holycow phoe: read great. thx!

12:42:50 shka this is quite exciting

12:42:57 phoe the implementation that can bootstrap itself off everything.

12:42:57 shka :-)

12:43:00 beach Now, I don't have any great hopes that existing implementation will actually ditch a single line of code in favor of the SICL equivalent. But then, a lot of what I have done so far has been publishable (because it is original work) and I intend to continue with new such things. So either way, I can justify my salary.

12:43:21 phoe "No matter what lisp you're running, you can be running SICL."

12:44:13 shka beach: well, it is reasonable thing to do

12:44:15 shka imho

12:44:29 shka for instance ecl is running thanks to jackdaniel

12:44:47 phoe beach: but I can see a possibility where implementations can simply make a tiny abstraction layer over their own modules, so they're "pluggable" and allow the user to choose e.g. what implementation of LOOP or the Lisp reader to use.

12:44:51 shka but he is just a single person, with other stuff to do

12:45:16 beach phoe: Maybe so.

12:46:06 shka if he could have less code to maintain, he may very well use SICL

12:46:50 beach Yes, jackdaniel has some plans to write a Cleavir-based compiler for ECL.

12:46:58 holycow oh, clus is common lisp ultra spec. neat!

12:47:44 shka phoe: btw, did you already have full standard in there?

12:47:49 shka *do you

12:47:53 beach Oh, another SICL invention is that of first-class global environments. I firmly believe that they are a requirement for a safe Common Lisp environment, and I believe such an environment is a requirement for a LispOS. But the recent discussion I read seems to go a completely different direction.

12:48:34 phoe shka: not yet.

12:48:42 shka right

12:49:33 beach When I hear the word "kernel" I space out, because (at least the way the word is typically used) I think the concept of a kernel is (or should be) a historical parenthesis.

12:50:22 presiden isn't kernel a mathematical term

12:50:48 beach Right, and "lisp" is a speech defect.

12:53:41 beach For my opinion on "kernels" see section 1.2.5 (PDF page 10, document page 6) of this document: http://metamodular.com/lispos.pdf

12:55:03 holycow holy crap. what? a lisp os spec?

12:55:21 holycow beach: amazing.

12:55:32 beach holycow: Where have you been? It's been around for a long time.

12:55:36 phoe holycow: beach is a very productive Lisper.

12:55:42 holycow clear with my head stuck up my arse

12:57:29 shka beach: we were discusing lispos on top of KVM today

12:57:43 beach I kind of read it, yes.

13:08:34 beach holycow: so when you wrote: "it would make more sense to generalize the [lisp] os out into userland and then write native interfaces into whatever kernel you want to target ...", I didn't quite understand what you meant.

13:09:14 holycow beach: i'm reading page 10 right now to understand your perspective on kernels

13:09:25 holycow i know that kernels do a lot more than 'just manage access to hardware'

13:10:12 holycow what i meant by that statement is that the most challengin part for any os is accessing hardware.

13:11:22 holycow why not let the linux kernel handle interacting with the hardware and let the lisp os handle the userland and beyond.

13:11:57 beach Sure.

13:11:58 holycow the naiive assumption being that you can either do this via cffi interfaces or perhaps by writing a kernel 'driver' that lisp can talk directly to

13:12:01 holycow however

13:12:11 holycow as i'm reading your kernel specs and history of why things are the way they are

13:12:24 holycow there are downsides as the linux kernel is a very special beast. as would ever kernel be.

13:12:36 p_l beach: what kind of technique did you use to speed up generic dispatch?

13:13:11 holycow ever = every

13:13:23 p_l well, kernel is technically a quite modern way of defining the OS, partially because of certain common characteristics which led to certain.... layout of code

13:13:23 phoe p_l: http://metamodular.com/generic-dispatch.pdf

13:13:34 p_l phoe: thx

13:13:38 beach p_l: Basically, avoid hashing and memory accesses. Number the classes, and compare small integers, some of which are constants: http://metamodular.com/generic-dispatch.pdf

13:14:28 p_l mhm

13:14:37 beach holycow: I can very well imagine using the Linux kernel to access the machine, which is why I think "bare metal" is a low priority. But I also don't want to use the abstractions proposed by the Linux kernel (everything is a byte stream, etc).

13:15:24 p_l beach: (un)fortunately "everything is a byte stream" is not that much common as one would think on linux...

13:15:56 p_l that said, doing it using L4 as core and having "hw servers" on linux might be much easier

13:15:59 p_l (and more lispier)

13:16:35 beach I also don't want a "file system" of course.

13:16:44 shka L4 is running on top KVM?

13:16:53 shka p_l: how that is supposed to work?

13:17:12 p_l shka: no, Linux running as userland process on L4

13:17:26 shka …i have no idea

13:17:31 holycow l4 is a microkernel

13:17:46 holycow or ... microkernel family? i don't know enough abou tit

13:17:58 shka and what linux is for in here?

13:17:59 p_l shka: there exists L4/Linux, which is Linux running on top of L4-family microkernel as userland process

13:18:08 shka but… why?

13:19:16 shka p_l: so L4 delegates hardware handling to linux?

13:19:21 p_l shka: usually it's done so that the Linux provides hw drivers but can be pre-empted by another process on same-level in hierarchy

13:19:21 holycow probably because microkernels have a security model built into them ... they are a bunch of servers talking to each other from what i understand

13:20:07 holycow and be isolated securely as well so if the driver krashes the microkernel does not. not sure how it applies to linux tho

13:20:35 p_l the linux kernel in systems based on L4/Linux runs at lower priority compared to "main" application

13:20:58 phoe it's no longer the core kernel, yo

13:21:31 phoe so basically L4 is a hypervisor

13:21:49 holycow ah, best desprition yet. right

13:21:54 phoe and can execute other servers next to Linux.

13:21:59 phoe AFAIU

13:22:53 p_l well, L4 delegates pretty much *everything* outside of scheduling and IPC to userland processes, including any kind of device access. Linux just runs as a process there and gets device-related resources delegated to itself

13:24:32 p_l beach: I strongly suspect that anything that is workable would, in the end, remind in implementation details a file system, even if not from top-level developer view. But I think this requires a nice whiteboard with colored markers and hour or two for discussion somewhere quiet :D (next ECLM/ELS?)

13:26:28 beach Deal!

13:28:06 p_l beach: an interesting quote from a filesystem developer: "Database people sometime get really surprised when it shows up that we talk same language and ask what database I'm working on." ;)

13:28:10 beach I personally just see RAM as a cache for disk (and other permanent storage).

13:28:34 shka well, yes

13:28:54 beach So I see the disk as being "the" Lisp heap.

13:29:01 p_l beach: There are many practical considerations to have a wildly different view of disk vs. ram, though - though it doesn't mean software and end user can't see it as one

13:29:34 beach As devices, they are of course very different.

13:29:43 p_l (including many chances for fun stuff to happen in the transforms involved between what's on-disk and what's in-memory)

13:30:47 beach That can be handled with check sums.

13:30:58 beach ... and probably should.

13:31:45 phoe beach: I wonder about garbage collection.

13:32:26 phoe If I simply get 4 GB of RAM and 40 GB of HDD swap space and call SBCL with 44G of heap, then it theoretically should work, right?

13:33:20 phoe Then it stores some of its data on hard drive, and then a full garbage collection is initiated.

13:33:54 phoe Won't the system slow to a halt because of all the swapping required by GC and happening because of a huge swap space?

13:36:14 holycow shka: beaches lisos specification of a lis operating system covers everything we chatted about this morning really

13:36:24 holycow if you haven't read it take a peak.

13:36:40 shka i already read it

13:36:43 beach p_l: How do existing systems handle this problem? Like what does Linux do to verify the transfer from "swap space" to RAM?

13:36:45 holycow thought so :)

13:37:09 TMA phoe: GC would need to be made aware of that

13:37:22 beach phoe: No, you would be able to use at most 40GB, the space of the disk.

13:37:47 shka this is not that complex, really

13:38:17 beach phoe: My estimation is that the vast majority of all data on a computer consists of videos and photos, i.e. data that is not required to be scanned by the GC.

13:38:20 shka basicly your ram is pool of fresh objects, HDD is pool of older objects that were swapped out

13:39:00 beach phoe: Such data would never be paged in, unless it is required by an application for viewing or listening.

13:40:10 shka anyway

13:40:31 beach The correct term is "paged out". The term "swapping" refers to moving entire processes between RAM and disk. But I know, it has been abused by Unix derivatives.

13:40:52 beach p_l: Did you see my question?

13:40:52 shka you can GC your persistant memory just once in a while

13:41:04 shka that would be the other p

13:41:09 shka phoe: ;-)

13:41:46 shka and further more, you can even GC persitant stuff in the background

13:42:29 shka not even that complex to implement

13:45:16 beach p_l: I frequently find myself in situations like this. I.e., I have some improvement to suggest. But then, suddenly, I am not only asked how that improvement works, but it is demanded that I improve on EVERYTHING.

13:45:19 beach And if I can't, I feel like my initial improvement is rejected. For example, recently, I talked about Second Climacs and its improved incremental parser. Suddenly, I was asked how it would handle a user-defined reader macro that forks the input stream and reads it later. I mean, I can't do that, but neither can Emacs or any other editor.

13:45:45 p_l beach: sorry, I had to unlock my account -_-

13:46:48 p_l beach: most operating systems don't checksum swap space at the moment directly in swapping code. Some systems do verification at lower layer in the stack, essentially guaranteeing that "this page I just copied from disk was not corrupted by outside influence"

13:46:57 beach phoe: There are some great new GC techniques around. For one thing, most GC invocations would be thread-local. Others would be incremental and concurrent.

13:48:04 beach p_l: Thanks.

13:48:43 p_l beach: I think a little bit of that is that us Lispers are just used to this crazy powerful thing, and when someone speaks of something novel we go "aha! now I could do something like this... could I actually do something like this?"

13:48:44 p_l ;)

13:49:04 holycow its also that all users suck

13:49:39 beach p_l: Possibly, but it gets very tiring and boring, at least to me. It makes me not want to discuss much of what I do. Which is why I most often don't.

13:49:47 p_l holycow: some users are just used to software sucking so much and being impossible to modify that they accept things and go on shoveling dirt chained :)

13:49:56 holycow i was once at a lug and some nice people were in there showing off some programming tricks they used to do some spam blocking and all the users started talking about this antivirus vs that solution

13:50:22 p_l beach: yeah. I try to not give out that vibe, but I am unfortunately a bit of "well-actually know-it-all" ^^;

13:50:31 holycow p_l: that too

13:50:51 beach p_l: This is not your problem of course. I am just explaining my reaction.

13:53:23 argoneus is there a clean way to write a function that returns whether a given list has an even amount of items?

13:53:33 argoneus I can't think of any way without using helper functions or variables

13:55:23 beach That would be clean.

13:55:47 p_l beach: I was generally imagining a much more complex system than just swapping to disk and back (even with checksums), because of practical reasons, distribution of code, etc.

13:56:20 p_l Using something like ZFS, but instead of emulating POSIX semantics on top use the underlying object graph more directly, with first-class *images* even

13:56:27 beach p_l: See, that's one of the problems I have not yet solved. Others would be eliminating wars and famine.

13:56:36 p_l haha

13:57:05 phoe argoneus: (defun oddp (x) (cond ((null x) t) (null (cdr x)) nil) (t (oddp (cddr x))))

13:57:10 phoe not really lispy because of TCO

13:57:20 holycow ACTION votes for zfs

13:57:30 p_l beach: I was more worried about getting an end result similar to how you install ITS from scratch on real hw, or avoiding cases like "how do you bootstrap CMUCL?????"

13:57:38 argoneus TCO?

13:58:14 beach p_l: Would it be OK if I let *you* worry about that?

13:58:47 p_l beach: oh, sure! It's fascinating thing, and in no way I'm arguing you should be the one to fix it

13:58:56 phoe s/TCO/TR/

13:58:58 phoe tail recursion

13:59:01 beach p_l: Thanks!

13:59:06 argoneus ah

13:59:11 beach argoneus: Do you just want to know whether the total length of the list is even or odd? Or are you counting a particular item?

13:59:12 p_l but if we had something generally hackable to start with, I'd be very tempted to spend some time hacking on it

13:59:19 argoneus beach: just true/false

13:59:20 p_l just that constant lack of time... -_-

13:59:44 beach argoneus: Are you counting ALL items in the list, or the occurrences of a particular item, like 234?

14:00:33 beach argoneus: First one: (oddp (length list))

14:00:48 beach argoneus: Second one: (oddp (count 234 list))

14:01:34 phoe argoneus: but I'd use a helper variable in this case if I was implementing such a function

14:02:52 beach p_l: I know that problem, alright.

14:04:05 beach p_l: Plus, I have come to realize that, even if we had money to throw at the problem, it would not help. All Lispers that are qualified enough to help out are already fully employed.

14:04:55 phoe beach: it might help them change their employment though

14:05:24 beach phoe: Yes, with enough money, that might work.

14:05:53 phoe if I had a stable job that would mean that I write Lisp, I wouldn't think thrice

14:06:04 p_l beach: I know lispers who would love a lisp-related job... just saying

14:06:09 phoe ^

14:06:10 beach phoe: I was thinking more of the type of money I could afford myself and that might be possible to get from crowdfunding.

14:06:12 shka well, i would also switch

14:06:18 p_l especially if we're talking CL, not Clojure

14:06:23 shka but i am not qualified for this

14:06:35 phoe beach: yes, I know. Crowdfunding is hard to call stable.

14:06:43 beach p_l: Oh, I do too. But the ones I know would not be able to do the job I would ask from them.

14:06:46 phoe Maybe soon, when I manage to get some financial buffer for myself.

14:07:25 p_l ACTION is on a consulting spree that unfortunately has very little lisp

14:07:32 p_l maybe I'll manage to put in some clojure

14:08:21 p_l most of what I do these days is system operations, not software dev

14:08:25 holycow beach: lots of patreon accounts make decent money

14:08:32 p_l holycow: not enough

14:09:05 holycow actually its probably on that end of the scale isn't it?

14:09:09 p_l that said, I should ask trinitr0n if he has one... man is doing Good Work by preserving Lisp Machines

14:09:36 holycow however, a set of subscription services might be some options

14:15:33 shka well, the problem here is that you need experts, but current experts are busy and future potential experts are getting expertise in other areas

14:15:47 shka because there is no room for non-experts in here

14:15:55 shka tadam!

14:15:58 shka ;-)

14:16:19 beach shka: so we need to come up with a way of turning non-experts into experts.

14:16:34 phoe boom, we're back at the topic of documentation.

14:16:46 beach That's part of it, yes.

14:17:19 phoe What are the other parts?

14:17:24 beach shka: Alternatively, we need an expert project planner who can extract individual tasks and suggest them to non-experts.

14:17:40 phoe Books? Tutorials? Videos?

14:19:17 beach phoe: Give me a few minutes to think of an answer to to that.

14:19:18 shka well, issue is that people need time and energy to learn

14:19:33 shka both are limited

14:19:46 shka especially if you need actual job to earn for a living

14:19:47 phoe beach: it's an important question. I value your input on it..

14:20:11 beach phoe: Good documentation makes it faster for someone who has already decided to become an expert to do so faster, and for someone who is already an expert to become even more productive.

14:20:37 shka phoe: documentation is like sex, you know…

14:20:40 phoe beach: So we know how documentation is important. What about other factors?

14:20:52 beach phoe: I think we need to find a way of turning people who don't know how to become experts into experts.

14:20:57 shka if it is good, it is really GOOOOOOOOOOD, and if not, it is better than nothing

14:21:05 phoe beach: Exactly.

14:21:26 beach phoe: That is a much harder problem. I don't know how to do it.

14:21:33 phoe I suspect it is going to a social process as much as a technical one.

14:21:40 beach phoe: ... and this is despite the fact that it is part of my job.

14:21:54 phoe beach: don't worry, we're all learning here.

14:22:04 p_l beach: I think you're better at that than me

14:22:12 phoe ^

14:22:18 beach p_l: You have very low standards. :)

14:22:22 p_l ACTION is stuck trying to teach his gf, and not because of her having problems learning...

14:22:28 p_l starting from basics is just *hard*

14:23:22 p_l especially when, like me, you want to do it right and thus you start looking towards at least skimming CS proper

14:24:30 beach There is also the problem of transmitting things that are not just knowledge, but also inspiration, desire to learn, estimating things (engineers acquire that over many years), etc.

14:25:27 beach phoe: Example: How much data is there on a typical desktop computer that is NOT just videos, sound, and text?

14:26:02 phoe beach: Such things would need to be transmitted through text, or so I expect.

14:26:03 beach ... and binary code.

14:26:13 phoe This is a thing I would like to read an article about.

14:26:28 p_l beach: surprising amount of databases, for one

14:26:29 phoe With some exercises later.

14:26:44 beach phoe: See, you are already determined to become an expert. So it is easy to generalize and think that everyone else is just like you. It just aint so.

14:26:53 p_l ^ exactly

14:27:38 beach phoe: And that's the mistake that nearly all professors make as well. They assume that the students are young versions of themselves. It just aint so.

14:28:07 shrdlu68 ACTION is almost done with cl-tls

14:28:54 p_l beach: another common problem especially *outside* academia is getting the basics of basics so you can go on to "interesting stuff"

14:28:56 phoe beach: I'll turn your question the other way - how possible it is for other people to become determined to become an expert?

14:29:32 phoe Because, okay, I am determined. Why? What are the causes? Can one influence that choice or the causes?

14:29:34 beach p_l: I am willing to believe you.

14:29:59 beach phoe: All good questions. I don't know.

14:30:18 beach phoe: The book I just read named "Peak" has some answers.

14:30:29 phoe beach: Link me please?

14:30:36 Reinisch I have a mental framework that might help in this line of inquiry.

14:30:38 beach p_l: Do you have a more concrete example?

14:30:48 beach phoe: Hold on...

14:30:52 phoe Reinisch: Link me please?

14:31:03 Reinisch Usually people want to solve a problem.

14:31:15 Reinisch phoe: no link, just an idea

14:31:34 phoe Reinisch: We have a problem - not enough experts. Can this problem count?

14:31:40 phoe s/Can/Does/

14:31:41 p_l beach: imagine you're trying to teach someone programming, from start. It's not academic environment, they are not students who are "by default" going to try to get the assignments done

14:31:48 shrdlu68 Not enough lisp experts?

14:32:01 Reinisch if you find someone that doesn't know how to add, and you try to teach them addition,

14:32:19 p_l beach: my issues were - how do I make it not entirely abstract at start? How do I prevent discouragement due to lack of visible progress?

14:32:28 Reinisch you may run into the problem of them not really being interested or invested in learning addition

14:32:51 beach phoe: https://www.amazon.ca/Peak-How-Master-Almost-Anything/dp/0670068764

14:32:57 Reinisch but if you take that same uninterested person and make them have a problem that addition will solve...

14:33:11 Reinisch like force them to be a grain counter for the village or what have you

14:33:29 Reinisch then when they have a hard time doing the thing they do every day (count)

14:33:37 p_l How do I balance my desire to give "proper" tools and education vs. what is readily applicable in workplace? Especially when the latter often involve byzantine smoke-and-mirrors boilerplate that isn't easy to explain to someone who just wants to program the computer to get something to happen

14:33:53 Reinisch suddenly you teaching them addition is amazing and helpful and they care and are invested.

14:34:31 beach p_l: Yes, I see. But you overestimate typical university students. They want to get a degree with as little effort as possible. They are not particularly interested in learning the material.

14:34:49 p_l beach: in a way, had we had a more programmable and inspectable computers, I'd have less of a problem - I'd show them how to play around in a REPL to bend the programs to do what they want

14:34:52 Reinisch so, to round-about answer phoe's question to beach: how possible it is for other people to become determined to become an expert?

14:35:16 Reinisch you have to first make it a problem that they aren't an expert

14:35:22 phoe Reinisch: so it's about motivation and leveraging peoples' frustration

14:35:24 shrdlu68 Reinisch: Not necessarily.

14:35:30 p_l beach: oh, I'm not overestimating them... with university students I just have the stick as available device, not just carrot

14:35:33 shrdlu68 People turn away from problems.

14:35:47 Reinisch well, a less antagonistic phrasing may help

14:35:49 phoe to have experts, you first need frustrated people

14:35:49 beach Reinisch: Well, professors who fail over and over again typically declare that "It is not possible to teach people who don't want to learn". It is a very comfortable opinion to have.

14:36:14 p_l beach: a student, if they want a degree, even as fast and easy as possible, gets to deal with exams and assignments etc. - if they don't do them, well, bye bye diploma

14:36:14 phoe ...which isn't as insane as it might first sound

14:36:24 shrdlu68 They go look for something else to do with their time.

14:36:36 beach p_l: Bah, cheating is much faster and requires less energy.

14:36:42 Reinisch usually people that want to learn magic tricks have the problem of not being able to confuse/surprise their friends

14:37:07 p_l (I'm probably a bit of outlier, as I entered university with huge desire to learn only to face non-academic problems breaking me down)

14:37:13 Reinisch in that scenario, there's less "frustration" and "having *problems*"

14:37:16 Reinisch but it'

14:37:20 beach p_l: That's exactly what I said to phoe.

14:37:28 Reinisch s still, I want to do a thing, so I must learn

14:37:43 p_l beach: Sometimes cheating works, but I don't think cheating on my university exams would have been easy or applicable

14:37:51 Reinisch beach: fair enough.

14:38:21 p_l you might cheat on assignments, maybe even on dissertation, but exams tended to be harder

14:38:23 beach p_l: You would be surprised. One time, I caught one student using his cellphone to take a photo of his exam and sent it to another student in the same room.

14:38:46 p_l beach: ah, that would be caught... quickly at my uni

14:38:58 p_l (by the very fact of how visible it would be)

14:39:25 Reinisch but the overall point I'm making is that all the questions I'm seeing about how to engage a potential student, or how to get through basics toward fun stuff...

14:40:02 Reinisch a lot of those questions seem more approachable (to me) through a framework of not trying to solve a problem that someone doesn't yet have

14:40:06 p_l and anyway, if I were a lecturer or worse at university, I wouldn't care about cheaters - I would care to find those who want to teach and let cheaters fall on their own cheats

14:41:12 Reinisch Good discussion/thoughts all around! cheers all

14:41:18 p_l I would not be invested in a cheating student out of 100 who fails exam where they suddenly have to think by themselves

14:42:11 p_l I would be invested when teaching someone 1:1 something they do want to learn, but they asked me precisely because just forcing through books and trial and error isn't appreciable use of time, anymore

14:42:17 beach Sure. Anyway, there seems to be a need, either to turn non-expert Lispers into expert Lispers, or to organize projects so that non-experts can contribute.

14:43:04 p_l and given that I personally learnt nearly everything through tinkering and trial and error and ancient dusty tomes speaking of FORTRAN77 as new language...

14:43:33 beach p_l: Yes, introspection. I know it well. It has very limited success when applied to others.

14:43:39 p_l (or of Common Lisp as recent standardization effort for an enterprise-class Lisp language)

14:44:25 beach In what context did you speak about Common Lisp that way?

14:44:56 beach Oh, this is again about extending the standard?

14:45:01 p_l beach: not myself, but a book I found as a small kid

14:45:07 beach Ah, OK.

14:45:19 beach Now I understand. Sorry.

14:45:35 p_l my first two contacts with Lisp involved a book about AutoLISP and another about Scheme (titled, actually, "The Programming Language LISP")

14:45:57 shka heh

14:46:07 discardedes book any good?

14:46:16 shka computer archeology :-)

14:46:16 discardedes i like to keey an eye out for good classics

14:47:16 discardedes i recently read Object Oriented Programming in CL by Sonya Keene, and I feel like I finally get OOP in a much better way

14:47:30 discardedes it was written in the 80s

14:47:43 discardedes learned more about OOP than years of using Java

14:47:50 beach discardedes: If you feel that way, I highly recommend you read the CLIM II spec. That's how I REALLY got it.

14:48:08 discardedes thanks for the recommendation!

14:48:16 beach Anytime.

14:48:25 discardedes :-)

14:48:37 beach http://bauhh.dyndns.org:8000/clim-spec/index.html

14:48:46 nyef Mmm. The CLIM II spec is dreadful in several respects, but in terms of its use of object oriented techniques it is very good.

14:48:58 discardedes ohhhh.... awsome!

14:49:00 beach I completely agree.

14:49:15 discardedes Im trying to figure out what to use OOP on

14:49:20 shka not really about lisp but Transaction Processing: Concepts and Techniques is surprisingly worth reading

14:49:24 discardedes i finally see it as something awsome and beautiful

14:49:29 discardedes something i could never see before

14:49:38 discardedes but its tricky to figure out how to apply it

14:49:51 discardedes i kinda feel you need a good grasp of the domain in order to apply it

14:49:57 discardedes and some good insights

14:50:42 discardedes thanks shka!

14:51:00 p_l discardedes: I think the book was pretty good, it was I think even a book written purely in polish, but I can't find it

14:51:24 discardedes oh.....

14:52:33 p_l ACTION goes to extreme and starts digging out national archives

14:52:56 p_l the book actually talked about Scheme, specifically T, but wasn't dogmatic about what is what in lisp ;)

14:53:05 discardedes Ive been working on modifying emacs in a way to suite how I want to work on it, and ive been thinking about how an interface design might be designed with oo technique

14:53:31 p_l huh, found some more polish books in archives

14:53:45 discardedes ha.... how hard to learn polish?

14:53:46 p_l for example "LISP: Programming for humanities"

14:53:50 discardedes :-)

14:53:56 discardedes really?

14:54:04 phoe p_l: I saw that book

14:54:08 phoe tiny and not really doog

14:54:13 discardedes even the title doesnt sound ... englih

14:54:14 p_l 1987

14:54:18 phoe discardedes: actually really hard.

14:54:29 p_l another is "Compiler for LISP 1.5 language"

14:54:34 phoe discardedes: "humanities" isn't a good translation I think

14:54:42 phoe more like, "LISP: programming for people studying humane arts"

14:54:51 discardedes ahhh....

14:54:54 phoe beach: so we have two paths we can take now.

14:55:00 discardedes that sounds like an interesting angle

14:55:01 phoe 1) how to turn people into experts

14:55:05 discardedes i love quirky stuff like that

14:55:07 discardedes :-?

14:55:10 discardedes :-D

14:55:11 phoe 2) how to make it possible for non-experts to contribute

14:55:39 p_l the LISP1.5 book is a report, 42 pages, from 1974

14:56:03 discardedes i think one way to create a culture would be to have some very user freindly social networking software developed for the culture

14:57:19 discardedes this might address non-expert contributions

14:57:56 p_l well, there are two big lisp communities outside of Scheme and CL or Clojure - Emacs Lisp and *AutoLISP*

14:58:24 p_l pretty sure most people who had contact with lisp in poland had because of AutoLISP

14:58:51 phoe and LISP 1.5

15:00:20 p_l phoe: I try to ignore the "shitty intro lecture at CS dept that everyone ignores because it talks about weird ancient languages that we won't ever use"

15:00:40 p_l where the lecturer uses a script that might be from 1970

15:01:42 p_l which leaves with people who remember few things like "lisp is old, slow, and written in all caps"

15:01:51 p_l (sorry for the rant)

15:02:07 p_l afk - "team building" thing at work -_-

15:06:40 beach phoe: Yes. I think number 2 is a possible one.

15:07:00 beach phoe: It might actually be a way to accomplish number 1.

15:07:24 phoe beach: I will want to have an audio talk with you, perhaps this evening.

15:07:27 discardedes beach: haha! The CLIM II spec is exactly the kind of stuff Ive been thinking about!

15:07:52 discardedes maybe its more in depth, but user interface type stuff

15:08:04 phoe I want to discuss a few things regarding both of these options in the context of what Reinisch said above.

15:08:45 beach phoe: Out of the question. In the evening I spend time with my (admittedly small) family.

15:08:56 beach discardedes: Glad you like it.

15:09:07 discardedes beach: thanks a million!

15:09:08 discardedes :-D

15:09:54 phoe beach: Okay! I will need to find some time during my work hours, then.

15:10:05 phoe Tomorrow morning perhaps?

15:11:06 beach Maybe so.