freenode/#lisp - IRC Chatlog
Search
11:36:29
flip214
White_Flame: My son is currently learning c# in school; but starting at that point he has no idea of all the (more or less hidden) lower complexities.
11:37:10
flip214
I think that starting from a low level and going up people can really appreciate all the layers inbetween - and end up with a rough idea where to look when something goes wrong.
11:39:06
Odin-
You may have a point about appreciation, but I think the real problem is when the lower levels are just completely ignored.
11:41:37
Odin-
Resulting in severe cases of "any sufficiently advanced technology" towards programming languages and (particularly) VMs.
11:44:36
Odin-
Well, I for one don't often bother with thinking about the operation of computers in terms of the quantum physics of semiconductors.
11:52:48
no-defun-allowed
Or whatever the counter thing is, I forgot. And it's nice that looking up "ABA problem" comes up with child torture methods.
11:57:50
no-defun-allowed
In my personal experience, going from Lisp back to Java at university made me want to transfer. So I did, and then I wanted to transfer again, but decided it was too hard to transfer once already.
12:16:44
contrapunctus
no-defun-allowed: I've been advised by at least three Lispers to write something in a non-Lisp for sake of employment...but the C family icks me out and doesn't really seem to offer anything very unique. Which leaves me with...ML family, Smalltalk, Erlang, Prolog, or Forth, so far ¯\_(ツ)_/¯
12:19:14
contrapunctus
Yeah, I'm thinking I have more fun as an amateur programmer anyway. Can pick whatever language or problem which tickles my fancy. Have better qualifications and experience in my other field, too.
18:13:57
stylewarning
Xach: I'm having difficulty repro'ing this QL report: http://report.quicklisp.org/2021-03-31/failure-report/magicl.html#magicl_ext-blas
18:14:47
phoe
stylewarning: how are you trying to reproduce it? it seems that this is not a compilation error but a FASL-loading error
18:15:19
phoe
so I guess that the way to reproduce it would be to compile everything, then try loading magicl/ext-blas in a clean image
18:17:42
phoe
hmmm, magicl/ext-blas depends on magicl/core which defines the magicl.backends package
18:20:03
phoe
my train of thoughts is, this is because it is mentioned in DEFTYPE and this type might be expanded in DEFMACRO WITH-BACKENDS... hmmmm, I'm thinking out loud
18:21:13
phoe
that's because DEFTYPE is like DEFMACRO, functions called by/vars accessed in the type expansion must be available in the compilation environment, hence EVAL-ALWAYS
18:24:06
stylewarning
phoe: I'm just going to commit that because it makes sense. I admit I haven't been testing on other platforms that are a bit more picky about *-time evaluation
18:25:32
phoe
I have no idea what is it about Xach-platform that makes it picky in this particular case
18:34:46
stylewarning
phoe: well based on your hunch I did https://github.com/rigetti/magicl/pull/133
19:21:58
gnUser
Hi guys, I am new to lisp (want to learn it for Emacs) any tutorial you can recommend? And anything cool to research that has been made using lisp?
19:23:00
phoe
asking because Emacs is written in Emacs Lisp, and #emacs is the best place to ask for that
19:23:31
minion
gnUser: please look at pcl: pcl-book: "Practical Common Lisp", an introduction to Common Lisp by Peter Seibel, available at http://www.gigamonkeys.com/book/ and in dead-tree form from Apress (as of 11 April 2005).
19:28:55
phoe
they are somewhat similar on the surface but the deeper you go, the more different they are
19:30:11
phoe
tl;dr elisp is a single-implementation unstandarized language mostly used for programming emacs whereas CL is a multi-implementation ANSI-standarized general purpose programming language
19:30:55
phoe
(I don't think the guile backend for elisp is ready enough to count it as a second implementation...)
19:31:27
phoe
of course you can use elisp for general programming, though it's kind of not what it is very commonly used for and what it is most optimized for
19:32:08
phoe
although CL does not have a programming editor of emacs quality and availability of software (yet). a lot of people actually use emacs as the editor to program in CL
19:32:27
phoe
with toolkits like slime+swank, or sly+slynk, written in both elisp (client) and CL (server)
20:20:47
jmercouris
anyone know who is responsible for this: https://github.com/Lisp-Stat/lisp-stat ?
20:48:45
jmercouris
Half the time when I read a bio I’m not sure if the author is delusional about their achievements or really IS something special
20:49:16
jmercouris
This applies more specifically to autobiographies and other documents of that nature
21:08:50
palter
Just wish Steve didn't grab the Symbolics username on GitHub. Would've liked to use it for the actual Symbolics software. (Just the VLM for now until JCMa can be convinced otherwise.)
21:10:52
palter
Can't release it because of ownership issues. But, Dave Schmidt can grant access to the VLM on a per-request basis.
21:11:50
palter
If you were a Symbolics customer, he'd probably say it's ok for me to give you access.
21:31:51
caoliver
palter on-going? I while back I purchased an XL1201 which wound up with memory issues. I really didn't get much value from the expenditure. I still have the Gen 8.3 CD, but no hardware anymore. Is there any chance I would qualify for VLM? I put about $3K into the box and maybe got to run it 24 times or so before I had problems.
21:49:39
Feldman
I mean it also depends on the country you are in, travel to Switzerland and pirate to your heart's content.
21:49:53
caoliver
OG 2.0 exists in a very public place though. I certainly didn't put it here, and I'm quite surprised it hasn't gone away.
21:50:44
caoliver
I do feel the principle interest is historical though, a museum you can run on your computer.
21:51:36
caoliver
So, I think the world is a far better place if OC, Medley, and Squeak are out there to poison childrens' brains. ;-P
21:52:14
caoliver
ACTION has Squeak and Medley on his box, though most of his hacking is in LuaJIT and C.
22:01:13
xristos
i'd rather do squeak and morphic these days than waste my time with the ever-changing mess that pharo has become
22:02:29
caoliver
Agree. It strikes me as far harder to hack on, and more than a bit over-engineered.
22:11:19
caoliver
I'll ask. My interest as you might guess from the scrollback is in no way commercial.
22:11:46
palter
The version in the wild is definitely not legitimate and is based on an emulator that has some serious problems.
22:12:13
caoliver
If I were doing that sort of thing, it'd be on one of the Linux native platforms such as SBCL or CCL.
22:12:59
palter
Open Genera (OG) is now up to 3.0. Genera is up to 9.0. Portable Genera which is e version for Intel and ARM is at 2.0.
22:18:23
caoliver
I should imagine aside from some hush-hush things, there really aren't many if any commercial customers.
22:19:00
White_Flame
or I guess I should ask, os Open Genera removed from the ivory instruction set now?
22:36:05
White_Flame
palter: the reason I'm curious is that I have my own from-scratch ivory emulator in CL, which also compiles ivory bytecoded functions into native CL functions.
22:37:25
White_Flame
but, I certainly don't have enough of the system platform guts to get any of the worlds that are out there on the internet to boot past a few million instructions
22:46:03
palter
Open Genera, now Portable Genera, still runs through an Ivory emulator. What’s changed is that the emulator is now native on both Intel and ARM64.
22:47:09
palter
And, yes, it runs on Apple silicon (i.e., the new Macs). In fact, supporting those systems is why we had to rebuild Genera from source. The ARM version already ran fine on Linux. (I have it on a Pi4 at home.)
22:48:03
White_Flame
the biggest tradeoff I was looking at was C-based can do MMU stuff to not have to emulate the virtual memory aspects of each and every memory access, but CL would give me JITting of the ivory functions
22:48:27
palter
But, Apple silicon changed the hardware page size to 16KB and Genera used 8KB on the VLM which meant that as soon as two adjacent Genera pages needed different protection settings, Genera would crash. GC would trigger that all the time.
22:50:29
White_Flame
from CL, I get around a lot of that by simply allocating the entire 40GB address space in RAM
22:50:45
caoliver
I hacked up a pthreads system for LuaJIT, but Apple decided to go their own way WRT IPC.
22:50:54
palter
I started working on the VLM again back in 2014 (or ‘15) to help Dave Schmidt with one of his clients. He needed a working Intel version of the emulator.
22:51:54
palter
We’ve also done a good deal of patches to Genera itself to better support ANSI CL. (We have a client running hunchentoot on Genera as his development platform)
22:52:08
no-defun-allowed
Does Portable Genera do any code generation, or does it interpret the Ivory code?
22:52:25
caoliver
I will confess that having a lispm that fits in an Altoids tin is a novelty I can believe in. ;-)
22:52:55
palter
As I remember, since I’m not at my computers right now, booting Genera and immediately shutting it down takes about 3 billion instructions. (I have traces.)
22:53:45
White_Flame
traces of correct execution are _exactly_ what I need to get unblocked from my development, but those world images would be different
22:55:10
palter
The Pi4 is a bit sluggish compared to my other systems but is still much faster than original Ivory hardware.
23:07:22
palter
A Pi4 is equivalent to an XL18300. But an Apple silicon Mac is equivalent to an XL103000.
23:15:22
caoliver
ACTION has no top-o-the-line stuff. My stuff is mainly surplus third gen core-i Optiplexes running a very slimmed down Slackware.