"Why no well-supported, modern, ML-style language that has a nice ecosystem/toolchain, and lets morons such as myself dip our toes in? Why they all gotta be like this?!"
Maybe it's on purpose and the goal is to filter out the unfit like a weeder class :P
Even though I only understood maybe 60% of this post it still made me chortle several times!
I'm not really a programmer or anything, but I am a Linux "daily driver" and (apologies if this comes out as too pop-eyed and evangelical... no wait where are you going....?) I struggle to understand why somebody would be trying to program in Windows, which must be like trying to swim in treacle, and I can't help feeling like all this stuff would be *sooo* much easier if you ran e.g. Windows-user-friendly Kubuntu inside a VM? Then you could install Haskell with one command (provided you don't mind "curl | bash" - in other words downloading a script off the internet, not reading it, and immediately executing it in a shell - not even remotely secure but ultimately no worse than downloading an .exe file off the internet...)
Or for F#, how good are AIs at coding in F#? I wonder whether, if they're good enough, you wouldn't need any textbooks because you could have a full-time personal teacher? (I would guess Claude would be the best AI for this, from what little I've used all the main ones...)
If I had to choose a programming language, I think probably I would choose on the basis of what the AIs were best at (which is probably a function of how much example code there is of each in their training corpus, and hence a function of how popular each language has been over the last 20 years or so..) just so's I could be taught by an AI instead of a textbook...
>I struggle to understand why somebody would be trying to program in Windows, which must be like trying to swim in treacle<
*cough* You're not the first I've heard say such a thing...
I've been *meaning* to try out a Linux partition or the like for many years now, and just never quite got around to it. (For a while, the issue was that I was always reading posts from people saying things like "help I can't get my router to work on Linux!" & responses along the lines of "okay, to fix this, you'll need to become very familiar with the instruction-set for your processor...", heh---but I think that we're far beyond that point, now, and for the most part everything will "just work" in Linux as it does in Windows. Or, well, *mostly* does in Windows--)
I like the look of Kubuntu over Ubuntu (KDE vs. Gnome?), and it seems pretty polished. Thanks for this; I'll give it a shot! Now perhaps I'll finally understand why all of the /real/ elite 10x devs do everything in Linux with shell commands.*
-------------
*(I have to admit: even just using PowerShell in Windows always makes me feel like a Real Programmer™---possibly even a 1337 h4x0r---regardless of whether I'm *actually doing* anything more technical than "navigating to new directory" or "creating new folder".)
--------------------------
>Or for F#, how good are AIs at coding in F#? I wonder whether, if they're good enough, you wouldn't need any textbooks because you could have a full-time personal teacher? (I would guess Claude would be the best AI for this, from what little I've used all the main ones...) [...] If I had to choose a programming language, I think probably I would choose on the basis of what the AIs were best at [...] {links}<
Thanks for this, too! It's a good idea (and I too have usually had the best luck with Claude, myself, for any semi-complex task).**
Y'know... I sometimes think about how difficult it must have been to just /find stuff out/, back before the 'Net: if you didn't have a textbook upon some topic, or back-issues of X or Y journal on the subject, you'd have to hope the local library did... and then there's the matter of just finding out if & where the particular question you had was actually touched upon, within all of those back-issues...
You can probably see where I'm going with this: your response here---pointing out that actually, I could probably just tell Claude "hey, how do I do X in F#? please show me & explain as you go along"---made me think of the similarities between "pre-vs-post Internet" & "pre-vs-post LLMs", heh. Perhaps we're seeing the beginning of a similar sea-change in ease-of-access to information (if that's even, really, a sufficiently broad descriptor for what AI is making possible!).
-------------
**(e.g. "compare & contrast the historical diets of (a) peasants vs. (b) nobles, at the following times-&-places: ..."---I've noticed that, say, ChatGPT will start contradicting itself without noticing, on this sort of thing, much more readily than will Claude; or at least, such was the case maybe ~half a year ago.
(I've tried to keep up on which model is doing the best by looking around on https://lmarena.ai/, but... well, do we trust the average normie to actually *accurately evaluate* the results they get? We maybe shouldn't, going by the correlation---or lack thereof---between my own perception & the results on the Arena... I'd be interested in your own evaluation, though, sometime!)
"Why no well-supported, modern, ML-style language that has a nice ecosystem/toolchain, and lets morons such as myself dip our toes in? Why they all gotta be like this?!"
Maybe it's on purpose and the goal is to filter out the unfit like a weeder class :P
...no! No, it cannot be, for then /I/ would be one of the unfit! [weeps angrily]
Even though I only understood maybe 60% of this post it still made me chortle several times!
I'm not really a programmer or anything, but I am a Linux "daily driver" and (apologies if this comes out as too pop-eyed and evangelical... no wait where are you going....?) I struggle to understand why somebody would be trying to program in Windows, which must be like trying to swim in treacle, and I can't help feeling like all this stuff would be *sooo* much easier if you ran e.g. Windows-user-friendly Kubuntu inside a VM? Then you could install Haskell with one command (provided you don't mind "curl | bash" - in other words downloading a script off the internet, not reading it, and immediately executing it in a shell - not even remotely secure but ultimately no worse than downloading an .exe file off the internet...)
Or for F#, how good are AIs at coding in F#? I wonder whether, if they're good enough, you wouldn't need any textbooks because you could have a full-time personal teacher? (I would guess Claude would be the best AI for this, from what little I've used all the main ones...)
If I had to choose a programming language, I think probably I would choose on the basis of what the AIs were best at (which is probably a function of how much example code there is of each in their training corpus, and hence a function of how popular each language has been over the last 20 years or so..) just so's I could be taught by an AI instead of a textbook...
References:
Kubuntu: https://kubuntu.org
VMware Workstation: https://blogs.vmware.com/cloud-foundation/2025/03/10/vmware-fusion-workstation-going-free-new-resources
Install Haskell on *buntu: curl --proto '=https' --tlsv1.2 -sSf https://get-ghcup.haskell.org | bash
Claude chat I had when writing the above: https://claude.ai/share/e201936c-4233-4b72-8e4e-e038609f1be9
>I struggle to understand why somebody would be trying to program in Windows, which must be like trying to swim in treacle<
*cough* You're not the first I've heard say such a thing...
I've been *meaning* to try out a Linux partition or the like for many years now, and just never quite got around to it. (For a while, the issue was that I was always reading posts from people saying things like "help I can't get my router to work on Linux!" & responses along the lines of "okay, to fix this, you'll need to become very familiar with the instruction-set for your processor...", heh---but I think that we're far beyond that point, now, and for the most part everything will "just work" in Linux as it does in Windows. Or, well, *mostly* does in Windows--)
I like the look of Kubuntu over Ubuntu (KDE vs. Gnome?), and it seems pretty polished. Thanks for this; I'll give it a shot! Now perhaps I'll finally understand why all of the /real/ elite 10x devs do everything in Linux with shell commands.*
-------------
*(I have to admit: even just using PowerShell in Windows always makes me feel like a Real Programmer™---possibly even a 1337 h4x0r---regardless of whether I'm *actually doing* anything more technical than "navigating to new directory" or "creating new folder".)
--------------------------
>Or for F#, how good are AIs at coding in F#? I wonder whether, if they're good enough, you wouldn't need any textbooks because you could have a full-time personal teacher? (I would guess Claude would be the best AI for this, from what little I've used all the main ones...) [...] If I had to choose a programming language, I think probably I would choose on the basis of what the AIs were best at [...] {links}<
Thanks for this, too! It's a good idea (and I too have usually had the best luck with Claude, myself, for any semi-complex task).**
Y'know... I sometimes think about how difficult it must have been to just /find stuff out/, back before the 'Net: if you didn't have a textbook upon some topic, or back-issues of X or Y journal on the subject, you'd have to hope the local library did... and then there's the matter of just finding out if & where the particular question you had was actually touched upon, within all of those back-issues...
You can probably see where I'm going with this: your response here---pointing out that actually, I could probably just tell Claude "hey, how do I do X in F#? please show me & explain as you go along"---made me think of the similarities between "pre-vs-post Internet" & "pre-vs-post LLMs", heh. Perhaps we're seeing the beginning of a similar sea-change in ease-of-access to information (if that's even, really, a sufficiently broad descriptor for what AI is making possible!).
-------------
**(e.g. "compare & contrast the historical diets of (a) peasants vs. (b) nobles, at the following times-&-places: ..."---I've noticed that, say, ChatGPT will start contradicting itself without noticing, on this sort of thing, much more readily than will Claude; or at least, such was the case maybe ~half a year ago.
(I've tried to keep up on which model is doing the best by looking around on https://lmarena.ai/, but... well, do we trust the average normie to actually *accurately evaluate* the results they get? We maybe shouldn't, going by the correlation---or lack thereof---between my own perception & the results on the Arena... I'd be interested in your own evaluation, though, sometime!)