<MichaelRaskin>
samueldr: how to prove that something went wrong with computers in a single picture? 1. that HDMI dummy. 2. And yes, it _is_ a useful product.
<Arahael>
MichaelRaskin: How do you use it?
<MichaelRaskin>
samueldr does — basically because nvidia is uncontrollable mess
<Arahael>
Could you elaborate?
<Arahael>
(Either of you! :) )
<Arahael>
I mean, why use an nvidia card if you run it headless?
<MichaelRaskin>
I would bet on CUDA/GPGPU
<Arahael>
Nivida still requires a monitor to do CUDA!?
<MichaelRaskin>
No, to do suspend
<MichaelRaskin>
Or maybe to do resume after suspend
<Arahael>
Wow.
<MichaelRaskin>
Despite what some people assume, even before the change of style people didn't get public expletives from Linus Torvalds without a reason
<MichaelRaskin>
It's eventually a static animation though, right?
<ldlework>
isn't there a nix thing which serializes the result of expression evaluation
<ldlework>
and you can ship it to another machine so the other machine doesn't actually have to execute/evaluate any expressions?
<MichaelRaskin>
Well, you can instantiate an expression then export the closure of the derivation
<ldlework>
that basically contains all the built binaries and stuff though right?
<MichaelRaskin>
No, you can get the derivation closure with only the derivations
<ldlework>
there is a user in emacs who is trying to make the case that we should give up executable expressions for a dead serializable format like json/toml/yaml etc
<ldlework>
because pushing executable scripts around is a security concern
<MichaelRaskin>
It does contain stored-in-nixpkgs patches, though (as they are necessary)
<ldlework>
those are just diffs thought right
<MichaelRaskin>
Emacs user. Does he use _any_ Elisp package from the web?
<MichaelRaskin>
Or she.
<MichaelRaskin>
Well, there are also some _scripts_ (*.sh) stored in Nixpkgs that are in derivation closure
<MichaelRaskin>
But re:evaluation — what exactly is the threat model, again?
<ldlework>
i guess you pwn nixpkgs or something no idea
<MichaelRaskin>
I want to instantiate a derivation, but I won't build it (fetchers have network access and can be malicious) — and all the pain in sansara be upon me if I actually run anything I obtain from Nixpkgs?
<ldlework>
they wont even cede the ground that it would mean incomprehensible maintainability losses
<ldlework>
so i'm not sure it's productive anyway
<MichaelRaskin>
So far I am trying to imagine a usecase for that passive-format version that wouldn't still amount to arbitrary code execution in case of attack
<ldlework>
I was more curious myself, if nix had an intermediate-format you could ship around that was already fully "computed" and just had to be ran
<ldlework>
right
<ldlework>
same with any other package manager too
<MichaelRaskin>
I mean, a format between user-run evaluation and daemon could at least be considered from the point of view of escalation
<ldlework>
i agree, if you mean that it too would be vulnerable to attack
<ldlework>
but it does remove vulnerabilities from the primary language itself
<MichaelRaskin>
Well, the daemon protocol doesn't use a Turing-complete language unquoted
<ldlework>
It'd be cool if gchristensen explained that to our friend (and for the benefit of #emacs as a whole)
<gchristensen>
what's up?
<emily>
i feel like recruiting people to join your argument in another channel is a bit...
<ldlework>
MichaelRaskin: I was at the point where i wanted to say "inb4 ansible-style yaml template extensions are suggested by a user"
<ldlework>
emily: they're already there
<gchristensen>
actually I wasn't :X
<ldlework>
not so much to rally to my side, but to help portray the reality about nix - but thx
<ldlework>
gchristensen: oh i've seen you there before
<gchristensen>
I hadn't rejoined yet
<MichaelRaskin>
ldlework: just remember you cannot convince someone, and you cannot usually win over the majority of bystanders, but you can recruit some bystanders to check out what you promote later
<ldlework>
again it's more "nix is X!" -.- no it's not "nix is Y!" no.. it's not. rather than "you're wrong, see why!" but i guess if no one cares about that neither do i :)
<MichaelRaskin>
I like Common Lisp and I don't find Emacs Lisp better than Vimscript, I definitely don't want to go to #emacs.
<gchristensen>
I enjoy their self-deprecating humor
<MichaelRaskin>
For a Vim'er Lisper to go to #emacs to argue with a regular is an enemy action
<gchristensen>
:)
kisik21 has quit [Remote host closed the connection]
kisik21 has joined #nixos-chat
hedning has quit [Remote host closed the connection]
kisik21 has quit [Remote host closed the connection]
johanot has joined #nixos-chat
johanot has quit [Quit: WeeChat 2.2]
johanot has joined #nixos-chat
<iqubic>
ldlework: How good are you at GO?
<ldlework>
not very
<ldlework>
I'm like 13k maybe
<iqubic>
I have no idea what that means.
* emily
~17k or something
<emily>
(discounting because of haven't played in ages...)
<iqubic>
I don't even understand the scoring rules in Go.
ekleog has quit [Quit: back soon]
<manveru>
iqubic: beginner is 25k, expert is 1k, basically :)
<__monty__>
Thing I dislike about go is how apparently it's impossible to explain scoring in simple terms.
<ldlework>
Any empty spot only able to reach a single color of stones is a good way to explain how to find what's territory.
<__monty__>
That'd require playing to the very end before scoring though.
<__monty__>
And aren't ambiguous situations still possible?
<ldlework>
I don't know what you mean by requring playing to the end.
<ldlework>
An empty spot able to reach both colors of stones is nobody's territory, at any point in the game.
<ldlework>
Not sure what you mean by ambiguous in this context.
<__monty__>
Playing to the end to determine any and all dead stones.
<ldlework>
Sorry I don't understand. There are rules for life and death, revolving around two eyes for a group. If a group has a large territory, it is up to the attacking player to determine whether they could invade and live inside (find two eyes of their own) or kill the group they're invading from the inside.
<ldlework>
It's not ambiguous, it's simply a judgement call. And it depends on how players actually perform during a fight. Mistakes can be made turning a reasonbly alive group into a dead one, or vice versa.
<ldlework>
It's a matter of pragmatics.
<__monty__>
All of that's too vague for two beginners to pick up the game.
<ldlework>
It's vague if you don't know anything about what I'm referring to but I've introduced a number of people to the game, and you have to explain what life and death is, but it's not "vague" beyond not already knowing everything about everything.
<ldlework>
To capture a group or stone in Go you must take every liberty of that group. A group with two eyes will have always have two liberties. This is because the opponent can only take one move on their turn. Trying to play in either eye one at a time is suicide since your own stone would have no liberties. Its just a fundamental part of learning the game.
<__monty__>
Scoring requiring a "judgement call", actually potentially multiple, is too vague for beginners.
<ldlework>
Scoring doesn't require a judgement call..
<__monty__>
Even recognizing an eye takes some practice. There's structure that look like an eye but aren't.
<ldlework>
Any open space that can only reach one color, is that color's territory.
<ldlework>
There's no ambiguity.
<__monty__>
The alternative is playing through to the end.
<ldlework>
Whether a territory can be successfully invaded is a judgement call.
<ldlework>
If you do not invade, your opponent's territory is theirs.
<ldlework>
If you do, and kill, it's yours.
<ldlework>
If you kill and fail, it's theirs.
<ldlework>
In each out come counting the result is not "vague" or ambiguous.
<__monty__>
Yes it is because it's not realistic to finish play on a full board.
<ldlework>
Sure it is.
<__monty__>
Not for beginners. Takes way too long.
<ldlework>
I don't understand your point.
<ldlework>
Is your point that go is literally ambiguous?
<ldlework>
Or that beginners have to spend time learning the game?
<__monty__>
No my point is it's too hard for beginners to pick up.
<ldlework>
Hence, nobody every learning how to play.
<ldlework>
Makes perfect sense.
<__monty__>
With chess you can literally do the entire explanation in like 20 lines.
<ldlework>
Yes and checkers even easier.
<ldlework>
Naughts and Crosses probably in less than 30 seconds
<__monty__>
I'm talking about the scenario where you have 2 beginners that are interested in learning together but don't have any experienced players to turn to.
<ldlework>
There's only the internet and thousands of introductory resources.
<__monty__>
Yeah, and the fact that's a *necessity* sucks imo.
<MichaelRaskin>
We can't have nice things, so people take Naughts and Crosses and create some inifinite-board rules that make recognising valid games NP-complete
<ldlework>
Learning the rules which allow for a valid move is actually easier than chess.
<ldlework>
Knowing what's good strategy is harder.
<ldlework>
MichaelRaskin: heh
<__monty__>
I'm not complaining about rules for moves. Though ko makes things more complicated than in chess imo.
<__monty__>
And I'm not talking about strategy at all.
<__monty__>
What I don't like is that scoring in go is such a nebulous topic.
<ldlework>
The simple rule for ko is you can't put the board back to how it was.
<ldlework>
And scoring is easy. Any spot that can only reach a single color, is that color's point.
<__monty__>
Scoring's not easy because you need to determine dead stones.
<ldlework>
Players agree on dead stones.
<__monty__>
And determining those requires playing on for beginners. And then when it's determined you have to go back to the point where you were counting.
<MichaelRaskin>
On a small board you can just play to completion, no?
<ldlework>
MichaelRaskin: yeah, and most beginners will start on 9x9 or even with capture Go
<ldlework>
Playing on 13x13 with my beginner friend right now
<__monty__>
I love the game *because* it's so different from chess. But I don't like how complicated scoring is, False eyes, komi (different board sizes), seki, dead stones.
<ldlework>
I think you're just calling the wider content of the game "scoring" and imo it isn't useful
<ldlework>
No two beginners can sit down and understand the whole of Go, nope.
<__monty__>
I'm literally talking about determining who wins.
<ldlework>
Yeah, doing that is easy.
<ldlework>
Learning when to stop attacking, or keep attacking is something else.
johanot has quit [Quit: WeeChat 2.2]
<__monty__>
You keep saying that but it's not. I've offered several reasons why I think so and unless you can satisfactorily explain all of them in a 5 line paragraph you're not changing my mind.
<ldlework>
I did though, you just ignored it. Counting/Scoring is based on what colors an empty space can reach.
<ldlework>
(the computer can tell you the score of the game at any point)
<ldlework>
could be a neat "hello world" for trying out a new programming language :)
<__monty__>
If a game requires a computer to determine the score that's a wart. You didn't go into how you know what's a fair komi, how to deal with false eyes or dead stones or how to determine/deal with seki.
<ldlework>
I'm saying that the rules for counting the score are so trivial that the computer can do it too
<ldlework>
And you could probably use it as a hello world programming challenge
<ldlework>
it is literally, for each empty space, walk to all other spaces without using diagonals
<ldlework>
if you can only reach one color of stones, you're a point of that color
<ldlework>
that's it
<__monty__>
Also, you need to count captured stones too, no? So your simple explanation's not even the entire story if we ignore all my gripes.
<ldlework>
everything else is strategy
<ldlework>
you both agree on the captured stones
<ldlework>
if someone has some tight position in the corner
<ldlework>
say, a bent four
<ldlework>
and neither player knows the vital point
<ldlework>
and the attacking player never invades
<ldlework>
and at the end of the game, neither player brings it up, the points go to the player with the bent four
<ldlework>
if there's just a stick-like group in one player's territory
<ldlework>
the computer will suggest them as dead stones since they don't surround any points
<ldlework>
but the players can realize this too
iqubic has quit [Ping timeout: 268 seconds]
<__monty__>
Dye realize how many words you're using to explain this? And you've still only touched on obvious bits.
<ldlework>
only because you're forcing the explanation of dead stones
<ldlework>
neither player will even know about this tactical point of the game
<ldlework>
they just need to know empty points that reach one color is a point for that color
<__monty__>
Because you *need* to know that to determine the score.
<ldlework>
no... you don't!
<ldlework>
i literally just described how it goes when neither player knows that they can invade further
<ldlework>
IE, strategy
<__monty__>
I'm done, have a nice night.
<ldlework>
o7
__monty__ has quit [Quit: leaving]
<ldlework>
beginner players of chess will not understand the concepts of hanging or pinning or counting trades
<ldlework>
not understanding trade counts is the same categorical thing as not knowing whether some territory can be successfully invaded or not
<ldlework>
you can expect players of either game to get into that on their first game with no teachers around
<ldlework>
can't* rather
<ldlework>
On my site I have a series called "cards" where I try to explain thing in the briefest manner while remaining effective: http://ldlework.com/projects/cards/
<ldlework>
I guess I have to do one on beginner go no