Logs: freenode/#haskell
| 2020-10-29 22:32:32 | <Rembane> | Because numpy and scipy |
| 2020-10-29 22:32:38 | → | GyroW joins (~GyroW@d54C03E98.access.telenet.be) |
| 2020-10-29 22:32:38 | × | GyroW quits (~GyroW@d54C03E98.access.telenet.be) (Changing host) |
| 2020-10-29 22:32:38 | → | GyroW joins (~GyroW@unaffiliated/gyrow) |
| 2020-10-29 22:33:07 | × | scasc quits (~szabi@213142096120.public.telering.at) (Remote host closed the connection) |
| 2020-10-29 22:33:24 | × | GyroW_ quits (~GyroW@unaffiliated/gyrow) (Ping timeout: 240 seconds) |
| 2020-10-29 22:33:33 | → | scasc joins (~szabi@213142096120.public.telering.at) |
| 2020-10-29 22:34:01 | <thblt> | And because you can't care about everything? Perfectionism is a curse. |
| 2020-10-29 22:34:10 | <koz_> | Also because argumentum ad serpentum makes people think Python is 'intuitive', 'natural' and 'easy to work with'. |
| 2020-10-29 22:34:20 | → | mananamenos joins (~mananamen@84.122.202.215.dyn.user.ono.com) |
| 2020-10-29 22:34:26 | <koz_> | None of the above are even remotely close to _testable_, much less true, but argumentum ad serpentum. |
| 2020-10-29 22:34:57 | → | conal joins (~conal@66.115.157.138) |
| 2020-10-29 22:35:22 | <thblt> | hahaha I love, and officially adopt, "argumentum ad serpentum" |
| 2020-10-29 22:35:32 | <Uniaika> | yeah I love it too |
| 2020-10-29 22:35:36 | × | Madars quits (~null@unaffiliated/madars) (Ping timeout: 272 seconds) |
| 2020-10-29 22:35:51 | <koz_> | thblt, Uniaika: Thank you, I'm glad my British private school education was not for nothing. |
| 2020-10-29 22:35:59 | <thblt> | (But well, testable or not, I honestly think it's true.) |
| 2020-10-29 22:36:45 | × | conal quits (~conal@66.115.157.138) (Client Quit) |
| 2020-10-29 22:36:52 | → | p-core joins (~Thunderbi@2001:718:1e03:5128:2ab7:7f35:48a1:8515) |
| 2020-10-29 22:36:53 | <thblt> | Highly dynamic languages are easy to get started on, although hard to write programs on that don't explode in mid-flight. |
| 2020-10-29 22:37:19 | → | conal joins (~conal@66.115.157.138) |
| 2020-10-29 22:37:20 | <koz_> | thblt: I would argue that this is _opposite_ of 'easy to work with'. |
| 2020-10-29 22:37:25 | <koz_> | Yes, you can write garbage easily. |
| 2020-10-29 22:37:30 | <koz_> | I don't consider this a virtue. |
| 2020-10-29 22:37:52 | × | conal quits (~conal@66.115.157.138) (Client Quit) |
| 2020-10-29 22:37:54 | <thblt> | koz_: my argument was about intuitive. I'm on #haskell, I'm not going to argue the virtues of dynamic typing. |
| 2020-10-29 22:37:59 | <thblt> | :) |
| 2020-10-29 22:38:19 | <koz_> | thblt: I dunno. Python grated on my intution badly even when I _wasn't_ a Haskeller. |
| 2020-10-29 22:38:28 | <texasmynsted> | But _why_ do people want to do ML in the first place? It would seem to me because it is useful. Think how much better it would be if the programs using the libraries were equally useful. |
| 2020-10-29 22:39:28 | × | britva quits (~britva@2a02:aa13:7240:2980:1ce4:625f:f093:45ef) (Quit: This computer has gone to sleep) |
| 2020-10-29 22:39:43 | <thblt> | koz_: to each their own, I guess. To me, Python is very, very obvious and my goto language for everything .1kslocs and below. |
| 2020-10-29 22:39:51 | → | heatsink joins (~heatsink@107-136-5-69.lightspeed.sntcca.sbcglobal.net) |
| 2020-10-29 22:39:55 | <texasmynsted> | I would think if one is willing to learn ML one would be willing to learn Haskell/Scala/WhateverLang. |
| 2020-10-29 22:40:04 | <thblt> | Also because of the stdlib let's be honest. |
| 2020-10-29 22:40:27 | texasmynsted | shrug |
| 2020-10-29 22:40:29 | <koz_> | thblt: Welcome to the 'it's not testable' part of my claim, and as far as I'm concerned, untestability means talking about whether it's true or not is meaningless. However, if you can get work done in Python, more power to you, because I just want to shower constantly in that language's presence. |
| 2020-10-29 22:41:18 | <thblt> | koz_: I'm not sure it's not testable, but I certainly can't be bothered into implementing a testing protocol :) |
| 2020-10-29 22:41:46 | <koz_> | thblt: You think you can come up with a scientific test for _intuitiveness_? Uhh, glhf with that. |
| 2020-10-29 22:41:58 | <texasmynsted> | Soon we will get to the testing vs types discussion . . . |
| 2020-10-29 22:42:15 | <koz_> | texasmynsted: Not the sort of testability I am referring to by a country mile. |
| 2020-10-29 22:42:17 | <thblt> | well, if it means something, it's testable. |
| 2020-10-29 22:42:49 | <koz_> | thblt: You think you can come with a _singular_ definition for intuitiveness? Uhh, also glhf with that. |
| 2020-10-29 22:42:56 | <dsal> | thblt: haskell is my goto for just about anything. |
| 2020-10-29 22:43:00 | <texasmynsted> | koz_: what are you referring to? |
| 2020-10-29 22:43:10 | <koz_> | texasmynsted: Testability in the scientific method sense. |
| 2020-10-29 22:43:21 | <koz_> | Not in the 'does my software launch nukes when I didn't ask it to' sense. |
| 2020-10-29 22:43:21 | <texasmynsted> | Python? |
| 2020-10-29 22:43:23 | <dsal> | I've seen a lot of people argue about, e.g., Haskell not being intuitive. It's usually more like, "Haskell doesn't match the mental model I've trained myself in for the last two decades." |
| 2020-10-29 22:43:25 | <texasmynsted> | Really? |
| 2020-10-29 22:43:48 | <koz_> | texasmynsted: Are you asking me if I really meant what I said? |
| 2020-10-29 22:43:50 | <koz_> | If so, then yes? |
| 2020-10-29 22:43:53 | <thblt> | dsal: I may get there, but the Prelude is an issue. For quick scripts, Python wins for me because of the batteries included aspect. |
| 2020-10-29 22:43:55 | <koz_> | Also what dsal said. |
| 2020-10-29 22:44:04 | × | heatsink quits (~heatsink@107-136-5-69.lightspeed.sntcca.sbcglobal.net) (Ping timeout: 240 seconds) |
| 2020-10-29 22:44:14 | <thblt> | dsal: isn't that exactly what an intuition is? |
| 2020-10-29 22:44:28 | <koz_> | Haskell is unashamedly different, and I like it for that. C is not such a good language that we should slavishly adopt its conventions. |
| 2020-10-29 22:44:32 | <dsal> | When you don't like python's batteries, you get to learn one of several competing things to make stuff work, and then your scripts don't work at some point in the future and you don't know how to recreate the environment. |
| 2020-10-29 22:44:48 | <dolio> | texasmynsted: I think there's nothing wrong with doing machine learning in Haskell or ML. I just think the people who gravitate toward machine learning would be the opposite of the kind of person who cares about the language it's done in being 'good'. |
| 2020-10-29 22:44:59 | <koz_> | And those batteries also suck in innumerable ways. Two of which I have encountered personally. As a _user_. |
| 2020-10-29 22:45:11 | <dsal> | thblt: It might be denotation, but connotation implies something closer to "natural." IMO, haskell is at least as natural as most other languages I've had to use. |
| 2020-10-29 22:45:18 | <koz_> | Python works if your buy-in to argumentum ad serpentum is high enough. |
| 2020-10-29 22:45:22 | → | GyroW_ joins (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) |
| 2020-10-29 22:45:22 | × | GyroW_ quits (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) (Changing host) |
| 2020-10-29 22:45:22 | → | GyroW_ joins (~GyroW@unaffiliated/gyrow) |
| 2020-10-29 22:45:27 | → | ensyde joins (~ensyde@99-185-235-117.lightspeed.chrlnc.sbcglobal.net) |
| 2020-10-29 22:45:43 | <koz_> | (by 'encountered' I mean 'in the last week or so') |
| 2020-10-29 22:45:59 | <dsal> | I used to use twisted quite a lot. Then I found someone who didn't like twisted made yet another thing to reinvent it and bragged about how much better and more intuitive it was (as long as you ignored the swaths of things I couldn't do with their framework). |
| 2020-10-29 22:46:01 | × | falafel_ quits (~falafel@71-34-132-121.clsp.qwest.net) (Ping timeout: 246 seconds) |
| 2020-10-29 22:46:03 | <koz_> | (and by 'user' I mean 'not a developer') |
| 2020-10-29 22:46:11 | <texasmynsted> | So testable like Coq or TLA+ or minizinc or agda, Idris, etc? |
| 2020-10-29 22:46:15 | <thblt> | koz_: Haha I believe you, big lib implies debatable choices. |
| 2020-10-29 22:46:28 | × | GyroW quits (~GyroW@unaffiliated/gyrow) (Ping timeout: 268 seconds) |
| 2020-10-29 22:46:40 | <dsal> | Python stdlib includes stuff like their async library they recommend nobody ever use, or the execution running stuff that they recommend nobody ever use. |
| 2020-10-29 22:46:40 | <texasmynsted> | maybe I just do not know enough about python |
| 2020-10-29 22:46:43 | <koz_> | thblt: So then 'batteries included' really means 'Big Daddy Guido did things the way he thinks best and you better like it'. |
| 2020-10-29 22:46:48 | × | Franciman quits (~francesco@host-82-54-10-114.retail.telecomitalia.it) (Quit: Leaving) |
| 2020-10-29 22:47:03 | <dsal> | Both are *almost* good, but missing just a bit of functionality. It's recommended you pick up something else external if you want to do any of these things that are included. |
| 2020-10-29 22:47:15 | <dsal> | For all intents and purposes, hackage is stdlib for my programs. |
| 2020-10-29 22:47:21 | <dolio> | texasmynsted: Because machine learning is the opposite of trying to clearly state the solution to a problem. :) |
| 2020-10-29 22:47:34 | <koz_> | dsal: That statement ports to any language and its hosted ecosystem, to be fair. |
| 2020-10-29 22:47:37 | <texasmynsted> | dolio: Ah ha! That makes sense. |
| 2020-10-29 22:47:46 | <koz_> | C.f. JS and npm, Python and pip, etc. |
| 2020-10-29 22:47:53 | <texasmynsted> | Okay I can understand that then. |
| 2020-10-29 22:47:53 | <thblt> | Hence my .1kslocs rule of thumbs. Anything above that, I can $package-manager-and-build-system init something. |
| 2020-10-29 22:47:57 | <dsal> | To be fair, python has more than one. |
| 2020-10-29 22:48:20 | <dsal> | You have to like, easy_install pip to get your virtualenv running. |
| 2020-10-29 22:48:43 | <dsal> | But yeah, once you get into the ecosystem, it kind of works. Some ecosystems operate more smoothly than others, though. |
| 2020-10-29 22:49:02 | <thblt> | I never use those. |
| 2020-10-29 22:49:03 | <texasmynsted> | I had installed anaconda ages ago. Now it tells me that my "free trial" is about to run out. No idea what it is talking about. |
| 2020-10-29 22:49:22 | <thblt> | (for python i mena) |
| 2020-10-29 22:49:24 | <thblt> | mean |
| 2020-10-29 22:49:35 | <merijn> | Python has like 30 package managers |
| 2020-10-29 22:49:40 | <merijn> | and their all god awful |
| 2020-10-29 22:49:54 | <koz_> | merijn: Very much this. |
| 2020-10-29 22:49:54 | <merijn> | I'd kill for cabal-install anytime I have to deal with pip and virtualenvs >.> |
| 2020-10-29 22:49:59 | <texasmynsted> | sure. Why should they work any better than the language its self? |
| 2020-10-29 22:50:04 | <thblt> | merijn: #emacs would find this reasonable. |
| 2020-10-29 22:50:10 | hackage | playlists 0.5.1 - Library and executable for working with playlist files. https://hackage.haskell.org/package/playlists-0.5.1 (PeterJones) |
| 2020-10-29 22:50:12 | <thblt> | :) |
| 2020-10-29 22:50:18 | <dolio> | texasmynsted: I think it's kind of like people who use matlab, although python is way better than matlab's language (last I checked). |
All times are in UTC.