←2023-02-06 2023-02-07 2023-02-08→ ↑2023 ↑all
00:11:25 <Sgeo> I tried to get ChatGPT to divide by 0. It started out better this time than the last time I tried. https://sharegpt.com/c/7z2BDJg
00:28:34 -!- immibis has joined.
00:28:58 -!- immibis_ has quit (Quit: HexChat).
02:22:15 -!- Lord_of_Life has quit (Ping timeout: 260 seconds).
02:22:43 -!- Lord_of_Life has joined.
02:45:36 -!- chiselfuse has quit (Remote host closed the connection).
02:45:54 -!- chiselfuse has joined.
03:12:44 -!- chiselfuse has quit (Remote host closed the connection).
03:12:51 -!- chiselfu1e has joined.
03:53:49 <zzo38> My brother agreed with me that the rule for legendary instants/sorceries is no good and that my unofficial rule for ongoing phenomena/dungeons/Sagas is good.
05:04:53 <Sgeo> No good in the sense that it doesn't quite match up with legendary permanents? Or how?
05:05:08 <Sgeo> I recently became willing to play MTG again. I don't think I am capable of staying away
05:36:07 -!- slavfox has quit (Quit: ZNC 1.8.2 - https://znc.in).
05:40:13 -!- slavfox has joined.
06:13:46 -!- bgs has joined.
06:31:06 <zzo38> Yes, it doesn't quite match up with legendary permanents. It means something different in this case so it isn't very good.
06:32:40 <zzo38> (The unofficial rule for ongoing though is more like the existing ongoing rule; it suppresses state-based actions that affect the object if it is not the source of any triggered ability or pending triggered ability.)
07:04:38 <esolangs> [[Queen]] https://esolangs.org/w/index.php?diff=106685&oldid=106684 * Pro465 * (+848) initial revision
07:05:50 -!- tromp has joined.
07:10:17 <zzo38> Can you make up any more Magic: the Gathering puzzles? I do not really intend to buy the cards and play the game, but I am interested in the puzzles. (And, if they published more books with them, then I might buy them, too; I do have one book)
07:15:15 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
07:37:59 <Sgeo> I'm not inventive enough to create puzzles, but I know there exists at least one website dedicated to MTG puzzles (possibilitystorm.com)
07:39:17 <zzo38> I have made some puzzles before, and I have some ideas but haven't made them
07:42:56 <zzo38> (One puzzle that I made up involves that you have to concede, in order to win. I have a few other ideas involving such a thing, too.)
07:45:42 -!- Sgeo has quit (Read error: Connection reset by peer).
07:50:31 -!- tromp has joined.
07:50:45 <b_jonas> Sgeo: I was wondering if the best way to play magic might just be to make a nostalgy format that uses old rules and old cards from the golden age of Magic. the bigger problem is, no two people will agree on when that was, everyone likes the time when they started playing and say that Magic was just getting worse after that. the smaller problem is that I don't have the Comprehensive rules and Oracle
07:50:51 <b_jonas> database going back that far in the past, the oldest I have is 2007-03 for cards and 2007-05 for rules.
07:51:08 <b_jonas> I think ideally I'd want between Dissension and Coldsnap
07:52:02 <esolangs> [[Queen]] https://esolangs.org/w/index.php?diff=106686&oldid=106685 * Pro465 * (+50) add cat example
07:53:48 <b_jonas> we could backport a few clearly good bugfixes from the future of course, such as the one bugfix that I'm responsible for back when Wizards was still running a forum and people working in Wizards read it; these days they closed the old style web forum and only communicate on Twitter or Snapchat or Tiktok or whatever the kids these days use
07:55:39 <b_jonas> (bugfix was for how the offering keyword from Betrayers of Kamigawa interacts with hybrid mana in mana costs, that was previously unspecified)
08:27:24 <esolangs> [[Counterfish]] M https://esolangs.org/w/index.php?diff=106687&oldid=103455 * Salpynx * (+4) clarify
08:27:41 <int-e> ^style
08:27:41 <fungot> Available: agora alice c64 ct darwin discworld elon* enron europarl ff7 fisher fungot homestuck ic irc iwcs jargon lovecraft nethack oots pa qwantz sms speeches ss wp ukparl youtube
08:28:01 <int-e> yeah not touching that
08:28:27 <int-e> Maybe if I wanted a horse, but I don't.
08:32:06 <esolangs> [[Counterfish]] M https://esolangs.org/w/index.php?diff=106688&oldid=106687 * Salpynx * (-4) /* Copy (duplicate and add) a prime encoded 'virtual' register */ Unicode exponents
08:45:12 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
08:52:57 -!- tromp has joined.
08:54:16 <esolangs> [[Queen]] https://esolangs.org/w/index.php?diff=106689&oldid=106686 * Pro465 * (+296) /* Examples */ add hello world program
09:18:09 <esolangs> [[Queen]] https://esolangs.org/w/index.php?diff=106690&oldid=106689 * Pro465 * (+90) add categories
09:20:25 <esolangs> [[Queen]] M https://esolangs.org/w/index.php?diff=106691&oldid=106690 * Pro465 * (+0) fix typo
09:55:12 -!- A_Dragon has quit (Ping timeout: 620 seconds).
10:07:36 -!- A_Dragon has joined.
10:11:19 -!- shachaf has quit (Read error: Connection reset by peer).
10:15:56 -!- shachaf has joined.
10:50:52 -!- __monty__ has joined.
10:58:08 -!- craigo has quit (Ping timeout: 246 seconds).
11:15:11 -!- wib_jonas has joined.
11:15:22 <wib_jonas> huh, you added a new elon style
11:17:32 <fizzie> Yeah, but it's not very good.
11:17:36 <fizzie> fungot: Are you a good CEO?
11:17:36 <fungot> fizzie: our best landing video to date, ai risk, our final invention of writing, is only 5500 years old, but as a weapon, so thats a reasonable question
11:19:03 <int-e> It's exactly 144 characters. I'm convinced.
11:19:19 <fizzie> I should have a better way of evaluating the goodness of a model than just generating 20 or so lines from it and trusting my feelings, then I could tweak the training parameters.
11:19:57 <fizzie> But I kind of suspect traditional language model performance metrics (perplexity, cross-entropy, ...) really measure how good it's for amusing people on IRC.
11:20:05 <fizzie> ^ don't
11:30:07 -!- SGautam has joined.
11:30:24 <wib_jonas> fizzie: I still think you should make a model based on lolcatbible.com . I may even have a very old dump of it, prepared to contain only the relevant text that I extracted, that I gave you at some point.
11:31:38 <wib_jonas> "traditional language model performance metrics" => also they only work well if the corpus is large enough that you can divide it to two really disjoint sets (including no retweets that quote the text of a previous tweet ideally), and I don't know if your corpus is large enough for that
11:40:38 -!- wib_jonas has quit (Quit: Client closed).
11:41:26 <fizzie> It was 16349 tweets, of which I used 15k for the training set and 1349 for the held-out set, which the language model training tool I use wants for tuning the Kneser-Ney discounting.
11:42:17 -!- wib_jonas has joined.
11:42:18 <wib_jonas> I see
11:42:42 <wib_jonas> so that's about two megabytes
11:44:19 <fizzie> To be honest, the Befunge implementation doesn't actually use the backoff probabilities of the generated model at all, and always picks the word from the highest-order n-gram set it can find for the current context. Which is just... incorrect, as far as the model goes.
11:46:10 <wib_jonas> well sure, but you don't want fungоt to be a really good language model anyway, that would lose his original personality
11:46:33 <fizzie> The manual for the tool I'm using says suitable size for the held-out set is "around 100 000 words/tokens", but I've only got 247309 in total.
11:48:40 -!- wib_jonas has quit (Quit: Client closed).
12:20:54 <esolangs> [[Special:Log/newusers]] create * Leol22 * New user account
12:21:49 <esolangs> [[Esolang:Introduce yourself]] https://esolangs.org/w/index.php?diff=106692&oldid=106662 * Leol22 * (+50) /* Introductions */
12:48:47 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
13:00:40 -!- wib_jonas has joined.
13:06:27 <esolangs> [[Language list]] https://esolangs.org/w/index.php?diff=106693&oldid=106663 * Pro465 * (+12) /* Q */ add Queen language
13:09:49 <wib_jonas> oh joy! python 3.11 changes the function datetime.datetime.fromisoformat so it now accepts more formats, including the short ISO 8601 formats that lack the hyphen and colon separators, such as "20230207" or "20230207T1309", and the week number formats like "2023-W06-2". so far this makes sense. but wait!
13:11:32 <wib_jonas> they also added some, uh, "reasonable extensions" to the formats that ISO 8601 allows, including that for the separator between the date and time, which should be "T" per ISO 8601, they now accept any one character. including a digit. which means dtm.datetime.fromisoformat("20230201013") is now parsed as if it was 2023-02-01T13, using the last zero
13:11:33 <wib_jonas> as the separator.
13:13:12 <wib_jonas> that's delightfully incompatible with perl's Date::Manip, which allows another "reasonable extension" which is omitting the separator between the date and time parts entirely, so  $d=Date::Manip::Date->new;$d->parse("20230201013"); parses that string as 2023-020T10:13, that is, day 20 in year 2023, after 10 o'clock, a different time. brilliant!
13:14:05 <wib_jonas> I side against python 3.11's datetime here, allowing a digit as the separator is stupid and error-prone, it will lead to at least silently missing errors.
13:14:20 <wib_jonas> sure, accept some alternate separators instead of "T" if you wish, but not digits please
13:16:48 <wib_jonas> Date::Manip is a bit overzealous in parsing so I'm not sure whether you do want to allow no separator at all, but that's an inheritence from the Date::Manip 5 days when the module's internal date format was something like "2023020713:10:00" (plus maybe some timezone thing)
13:28:13 <wib_jonas> .oO(there are fifteen different timestamp parsing modules with all different semantics. I know! I'll just make one timestamp parsing module that has the best semantics covering all use cases, and release it as free software!)
13:39:29 -!- SGautam has quit (Quit: Connection closed for inactivity).
13:43:30 -!- craigo has joined.
13:50:55 -!- craigo has quit (Quit: Leaving).
14:02:08 -!- tromp has joined.
14:14:57 -!- Lymia has joined.
14:34:08 -!- perlbot has quit (Ping timeout: 248 seconds).
14:34:11 -!- simcop2387_ has joined.
14:34:27 -!- simcop2387 has quit (Ping timeout: 252 seconds).
14:35:31 -!- simcop2387_ has changed nick to simcop2387.
14:38:10 -!- perlbot has joined.
14:41:54 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; $d->parse("20230302122") or print $d->printf("%O = %KT%X\n"); # so I understand this one, it parses as year 2023, day of year 030, hour 21, minute 22
14:41:55 <HackEso> 2023-01-30T21:22:00 = 2023-030T21:22:00
14:42:09 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; $d->parse("20230302922") or print $d->printf("%O = %KT%X\n"); # but where the heck does this come from?
14:42:11 <HackEso> 2664-03-05T15:17:33 = 2664-065T15:17:33
14:42:47 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; print $d->version(1); # for the record
14:42:48 <HackEso> 6.76 [utc]
14:56:05 <fizzie> 20230302922 seconds since January 1, 1970?
14:56:09 <fizzie> `` date --date=@20230302922
14:56:49 <HackEso> Mon Jan 28 00:35:22 UTC 2611
14:57:01 <fizzie> Hmm. Well, right century, but not quite.
14:57:37 <wib_jonas> fizzie: I was thinking of that, but it doesn't seem like it. for that, Date::Manip wants the format "epoch 20230302922", and in any case the numbers just don't match
14:58:03 <fizzie> `` date --date=@$(( $(date +%s) + 20230302922 ))
14:58:04 <HackEso> Sat Mar 5 15:33:26 UTC 2664
14:58:08 <wib_jonas> `dateu @20230302922
14:58:08 <HackEso> 2611-01-28 00:35:22.000 +0000 UTC January 28 Monday 2611-W05-1
14:58:14 <wib_jonas> this is what I use by the way
14:58:19 <fizzie> 20230302922 seconds after the current time seems to be a pretty good match ^
14:58:27 <wib_jonas> ``` date +%s -d "@20230302922 "
14:58:28 <HackEso> 20230302922
14:59:22 <wib_jonas> huh...
14:59:45 <wib_jonas> `dateu now +202303022922sec
14:59:47 <HackEso> 8433-10-30 19:08:28.528 +0000 UTC October 30 Sunday 8433-W43-7
15:00:01 <wib_jonas> `dateu now +20230302292sec
15:00:02 <HackEso> 2664-03-05 15:24:54.125 +0000 UTC March 5 Saturday 2664-W09-6
15:00:06 <wib_jonas> `dateu now +20230302922sec
15:00:07 <HackEso> 2664-03-05 15:35:29.077 +0000 UTC March 5 Saturday 2664-W09-6
15:00:11 <wib_jonas> that, yes
15:00:15 <wib_jonas> odd
15:00:23 -!- iovoid has quit (Ping timeout: 255 seconds).
15:00:40 -!- moony1 has joined.
15:01:22 -!- Bowserinator has joined.
15:01:27 -!- iovoid has joined.
15:02:10 -!- moony1 has changed nick to moony.
15:03:10 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; $d->parse("20230302922") and die; $n = $d->new("now"); print $d->printf("%O; "), $n->printf("%O; "), $n->calc($d)->printf("%sys; ");
15:03:12 <HackEso> 2664-03-05T15:38:33; 2023-02-07T15:03:11; 20230302922;
15:03:32 <fizzie> > Most valid deltas can be used to specify a date, and the date is defined as that delta added to "now". Refer to the Date::Manip::Delta documentation for a list of valid delta formats.
15:03:33 <lambdabot> <hint>:1:48: error: parse error on input ‘,’
15:03:57 <fizzie> Should remember not to use > for quoting here.
15:03:59 <wib_jonas> indeed it's the number of seconds from now. that's odd. I'll check on a later version of Date::Manip and perl later, just to be sure, and if it still does that, I'll ask Sbeck
15:04:39 <fizzie> I think the logic is that it's a valid delta (a Date::Manip::Delta string can omit the word "seconds" and have just the one field), so that's how it gets parsed as.
15:05:08 <wib_jonas> sure, but it's still inconsistent with how 20230302122 is parsed
15:06:37 <wib_jonas> also this is funny because now 20230302122 can mean three different times: 2023-030T21:22 (Date::Manip's preference), now plus 20230302122 seconds, or 2023-03-02T22:00
15:06:59 <wib_jonas> though I still think the last one, python 3.11's interpretation, is just a straight up bug
15:08:36 <wib_jonas> this is now definitely esoteric territory by the way
15:10:04 <fizzie> Mhm, right. Well, it can't parse 20230302922 the same way it parsed 20230302122 (because it'd be hour 29), so it... falls back to interpreting it as a delta? Something like that. Sounds like one of these things where the implementation is the specification.
15:16:11 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; $n = $d->new("now"); for $s (20230228..20230232) { print"$s => "; if ($d->parse($s)) { print "parse error: ", $d->err, "\n"; } else { print $d->printf("%O = %KT%X = "), $n->calc($d)->printf("now+%syssec\n"); } }
15:16:13 <HackEso> 20230228 => 2023-02-28T00:00:00 = 2023-059T00:00:00 = now+1759428sec \ 20230229 => parse error: [parse] Invalid date \ 20230230 => parse error: [parse] Invalid date \ 20230231 => parse error: [parse] Invalid date \ 20230232 => 2023-09-29T18:46:45 = 2023-272T18:46:45 = now+20230233sec
15:16:26 <wib_jonas> ^ delightful, it occurs even for 8-digit and 10-digit strings
15:17:14 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new; $n = $d->new("now"); for $s (20230430..20230432) { print"$s => "; if ($d->parse($s)) { print "parse error: ", $d->err, "\n"; } else { print $d->printf("%O = %KT%X = "), $n->calc($d)->printf("now+%syssec\n"); } }
15:17:16 <HackEso> 20230430 => 2023-04-30T00:00:00 = 2023-120T00:00:00 = now+7029765sec \ 20230431 => parse error: [parse] Invalid date \ 20230432 => 2023-09-29T18:51:07 = 2023-272T18:51:07 = now+20230432sec
15:18:08 <wib_jonas> 20230430 is a valid date meaning 2023-04-30 as exected; 20230431 is invalid because 2023-04 does not have a month 31, and 2023-04-32 means now +20230432 seconds
15:23:03 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new(); $n = $d->new("now"); $s = "201612312359"; for $_r (0..2) { $u = "$s UTC"; print"$u => "; if ($d->parse($u)) { print "parse error: ", $d->err, "\n"; } else { print $d->printf("%O = "), $n->calc($d)->printf("now+%syssec\n"); } $s++ }
15:23:05 <HackEso> 201612312359 UTC => 2016-12-31T23:59:00 = now+-192554644sec \ 201612312360 UTC => 8411-12-11T11:29:04 = now+201612312360sec \ 201612312361 UTC => 8411-12-11T11:29:05 = now+201612312361sec
15:24:45 <wib_jonas> ^ that's got to be a bug, right? 2016-12-31T23:60 UTC was a real leap second, so even if you don't handle leap seconds, 201612312360 should be parsed correctly as some time around 2016-12-31T23:60, right? assuming, as Date::Manip documents, that the T can be omitted
15:25:21 <wib_jonas> or hmm
15:25:27 <wib_jonas> that would be a leap minute, which is invalid
15:25:37 <wib_jonas> `perl -weuse Date::Manip::Date; $d = Date::Manip::Date->new(); $n = $d->new("now"); $s = "20161231235959"; for $_r (0..2) { $u = "$s UTC"; print"$u => "; if ($d->parse($u)) { print "parse error: ", $d->err, "\n"; } else { print $d->printf("%O = "), $n->calc($d)->printf("now+%syssec\n"); } $s++ }
15:25:39 <HackEso> 20161231235959 UTC => 2016-12-31T23:59:59 = now+-192554739sec \ 20161231235960 UTC => parse error: [parse] Delta too large \ 20161231235961 UTC => parse error: [parse] Delta too large
15:26:06 <wib_jonas> but even 20161231235960 UTC, which should be a leap second, isn't parsed right
15:28:09 <wib_jonas> well, the previous thing still applies. I'll have to test with a later version of Date::Manip, and unless that clears up everything (unlikely based on the Changes documents) I'll have to mail Sbeck
15:42:15 <wib_jonas> I mean even if it's not a bug, it should be documented as a caveat you have to be careful about, or in the changelogs as fixed
15:48:19 -!- Sgeo has joined.
15:57:02 -!- wib_jonas has quit (Quit: Client closed).
16:02:08 <esolangs> [[Queen]] https://esolangs.org/w/index.php?diff=106694&oldid=106691 * Pro465 * (+129) /* Examples */ add truth machine program
16:27:54 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
16:37:49 -!- tromp has joined.
16:42:46 <esolangs> [[Binary Brainfuck]] https://esolangs.org/w/index.php?diff=106695&oldid=106682 * Bbf * (+294) /* Commands */
16:43:19 <esolangs> [[Talk:Binary Brainfuck]] https://esolangs.org/w/index.php?diff=106696&oldid=106666 * Bbf * (-79) Blanked the page
16:45:32 <esolangs> [[Binary Brainfuck]] https://esolangs.org/w/index.php?diff=106697&oldid=106695 * Bbf * (+13) /* Hello World! Script */
16:47:02 <esolangs> [[Binary Brainfuck]] https://esolangs.org/w/index.php?diff=106698&oldid=106697 * Bbf * (+2) /* Hello World! Script */
17:18:01 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
17:21:42 <esolangs> [[Binary Brainfuck]] https://esolangs.org/w/index.php?diff=106699&oldid=106698 * Bbf * (+113) /* Hello World! Script */
17:34:23 <zzo38> We could try to work to write rules for playing a old style game if wanted, with some newer corrections too. I do have some ideas about how this could be done, although it would also require writing errata for some cards
18:09:47 <esolangs> [[TurtleDigits]] https://esolangs.org/w/index.php?diff=106700&oldid=94927 * NikiTricky * (+74) Add more categories
18:15:54 <esolangs> [[Listack]] M https://esolangs.org/w/index.php?diff=106701&oldid=106677 * McChuck * (+0) /* Example programs */
18:37:12 -!- tromp has joined.
19:37:06 <esolangs> [[PRINT/Concept Interpreter]] https://esolangs.org/w/index.php?diff=106702&oldid=106631 * SpaceByte * (-16)
21:04:51 -!- bgs has quit (Remote host closed the connection).
21:44:06 <b_jonas> fizzie: wait, it gets better. Date::Manip parses 111111-11 as meaning 2011-11-11T11:00, which is documented. but 111111-55 is parsed as 2011-11-11T??:55 where the hour is taken from the current date but the minute is overwritten with 55.
21:44:40 <b_jonas> so I turned that to https://www.perlmonks.com/?node_id=11150225
21:51:06 <esolangs> [[Brainfuck Encoded Concatenative Calculus]] M https://esolangs.org/w/index.php?diff=106703&oldid=103166 * Olus2000 * (+30) /* Computational class */ typo in `cat call` reduction
22:51:28 <esolangs> [[Falsish]] M https://esolangs.org/w/index.php?diff=106704&oldid=104788 * McChuck * (+14) /* See also */
22:52:23 <esolangs> [[Listack]] M https://esolangs.org/w/index.php?diff=106705&oldid=106701 * McChuck * (+24) /* Features */
23:19:31 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
23:30:39 -!- __monty__ has quit (Quit: leaving).
←2023-02-06 2023-02-07 2023-02-08→ ↑2023 ↑all