Thanks for this. I enjoy reviewing your videos each year. I notice you mention that a Map is just a .net Dictionary. This isn't actually true, as it has Log n lookup time instead of constant lookup time. This is because Set and Map are both tree data structures, which helps with immutability and persistence. I've been using Sets and Maps to speed up my code, but I might have to think again about this!

Glad to see you're back for 2024!

I got pretty good time processing the list forward using binary recursion (or I guess ternary for part 2) at 0.15s for part2. At each step you can check if the sum or the product would be greater than the target and stop checking down that branch. However, it seems processing backwards allows for pruning the branches better. Glad I checked your video out. (code example is below in read more) let isEquationSatisfiableWithConcat (lhs, calibrations) = let rec loop runningTotal calibrations = match calibrations with | [] -> runningTotal = lhs | c :: remainingCalibrations -> let runningTotalProduct = runningTotal * c let runningTotalSum = runningTotal + c let runningTotalConcat = concat runningTotal c let productSatisfies = if runningTotalProduct > lhs then false else loop runningTotalProduct remainingCalibrations let sumSatisfies = if runningTotalSum > lhs || productSatisfies then false else loop runningTotalSum remainingCalibrations let concatSatisfies = if runningTotalConcat > lhs || productSatisfies || sumSatisfies then false else loop runningTotalConcat remainingCalibrations productSatisfies || sumSatisfies || concatSatisfies match calibrations with | c :: cs -> loop c cs | _ -> failwith "Expected at least one calibration component"

Thanks for posting, I am new to F# and I learn so much from your videos. I solved this puzzle by using a quicksort using a rule checker function to portition the update list. If the sorted list matches the initial list then it is valid. For part 2 the sorted list is the one we want. Not very efficient but my head was scrambled by over complicating things!

On part2 couldn't you keep the parser from part1 and insert a mutually recursive function before the regex from part1. Something like `let rec splitOnDont enabled remainder = if remainder = "" then enabled else let [|e;r|] = lines.Split("don't()", 2) splitOnDo (enabled + e) remainder and splitOnDo enabled remainder = if remainder = "" then enabled else let [|_;r|] = lines.Split("do()", 2) splitOnDont enabled remainder`

Learning more F# and trying to be idiomatic as I go along and check those videos out at the end of the day. I thought I was being really smart using fold and keeping state of 2 digits+the current one, but the second part really screwed me over with a bunch of edge-casing and all.

Loving the videos. Will you be continuing with your Web Api series? im invested!

Rider seems nice. I installed it after the licencing change to free for non-commercial use. But I still spend more time in VS Code because I am lazy to set it properly and I am missing some hints provided by Ionide. BTW: do you prefer in part1 List.zip |> List map over just List.map2 because of readability or some other reason?

Thanks for doing these! Unlike previous years, I've forced myself to do the puzzle first, which makes watching your version even more valuable. List.transpose is cool!

First time I saw a non-responsive AoC page! Good on you for livecoding this! <3

I had totally forgotten that List.transpose exists and messily coded it by hand. Thank you for the reminder!

Good to see you back here on KZbin. I really enjoy your F# live coding.

Thanks for doing these videos! It's interesting to see the though process of someone who's more familiar with the language. Probably I will do this year in Scala instead of F#, but still, it's really great to see that there's content available for both languages! Are you planning on uploading this year too? :)

YES YES YES! This is exactly what I needed! This series is a gem I hope this helps further the adoption of F# by mainstream developers

Would love to see episode 3!!

You know what would probably be more interesting instead of googling "how do i create a record?"? Ask codeium right there in your IDE.

I switched from VS to Rider not too long ago. Once I got over the initial Who Moved My Cheese friction, I simply could not believe the quality of life improvement I was getting. It's embarrassing at this point how much better the F# experience is within a JB product than within a MS one. That said, my employer paid for the license and I have no idea of the cost because I never looked into it. I do think I'd struggle to go back to VS at this point, and this recording has reminded me of the day-to-day struggles with doing F# in it.

Will this series cover authentication and authorization?

Besides from demonstrating some cool F# features, thanks Jo for confirming VS is still the lesser IDE 😀

Thank you for keeping your stuff-ups in the video--it's so reassuring to see someone making the same mistakes I do--installing Nuget packages into the wrong project (and fix by copying and pasting project config) or getting backslashes and forward slashes muddled up. I suppose that almost everyone does this kind of thing, but it's so rare to actually see it--made me feel much better about myself! Will definitely be following along--looking forward to more (especially persistence and data-validation). :)

I think Giraffe got new maintainer(s) so it should be ok. You're making some great content! Thanks!

keep going, really nice content!

I've tried Rider and in many places seems to be more friendly to F# than Visual Studio. It will autocomplete record fields for you for example, which to this day VS still won't do. It has had this nasty bug for a long time where indent guides are not correctly aligned, which is a big problem when working with whitespace-sensitive languages :D. There's also the issue of licensing so ... By the way, how do you get code autoformatting in Visual Studio? 🙂 Oh, I see there's a VS extension for Fantomas, it's probably that.

Good stuff! Funny how you start on the "outside" of the application, I always begin at the domain side :-)

Finally! Something real world style in F#.

Good stuff! Funny to see (or maybe not) that VS is still acting strangely with the red squiggly lines - I remember that from your AoC videos! Must be frustrating as it slows down the coding. Do you have the same problem in VS Code?

Keep making these videos, they are a lot of fun.

I’ve been looking forward to this

Looking forward to the next episode. Any thoughts on Giraffe's current status and its most recent fork, Oxpecker?

Perfect timing! Planning on spiking out this exact thing with a few of the team this week!

Thanks for introducing me to Unquote! I'm really looking forward to further installments in this series!

This is awesome! Excited to build with F#

This is just the foothold I need to get started making projects in F#. Thank you so much for this.

Pretty cool. More F#!

Just stumbled on property testing today, I've written a lot of algorithmic code that would have benefited from it. If the random data causes a test failure, we'd still need TDD developed tests to find the problem as the next run may pass. You were also doing manual mutation testing to prove that your property tests were valid. Could be a powerful combination where it matters. For day-to-day testing, I'd never get anything done with this as the test turnaround would be so slow. Maybe, segregated tests would work. Very interesting topic, thanks for the demo.

Very cool using ray casting to help solve this problem - makes me think of Wolfenstein and Doom tech.

Very nice i liked it a lot. I got a question on how do u execute a specific Method and execute it?

is it possible to pay you to help me with a simple fsharp programming assignment? Estimated time is maybe 1 hour

First part of this day is really simple and clear, especially if compared to a few previous days. Thanks for the video!

I've been learning F# for the past month and found working on AoC alongside you very productive. First I do my best at the problem, then look at your vids to see what language features I could use to have a cleaner solution. Thanks for putting them out!

I really liked this series so far. Your videos made me want to give F# another spin after only dabbling with it in the past.

What kind of extension are you using for auto-formating?

Your idea with Lenth was good, it should work if you used length-1 because afaikt there can Veblen just one overlap

It’s good that you found the bug 😊 they do have an example in The test, eightwosometving with the t overlapping. It’s how I spotted it myself.

In handType you could have counted the Js and enrich the pattern match tuple. Then a (IsFourOfKind, 0) is still a FourOfKind but a (IsFourOfKind, 1) becomes a FiveOfKind for example. :)

If I understood your call for help on remembering compare results, then I remember the compare function as subtraction. So compare a b is like a - b. If the result is positive then a is greater than b. Negative means that a is less than b. Zero then they are equal.

I used groupby on list of cards in a hand. In addition I kept the max count of cards in the groups. This way it was easy to match with hit count. Like this let toQuality1 hand = let hitCount, maxEqualCardsCount = groupCards hand match hitCount with | 1 -> Five | 2 -> //Four, House match maxEqualCardsCount with | 4 -> Four | _ -> House | 3 -> // Three, TwoPair match maxEqualCardsCount with | 3 -> Three | _ -> TwoPairs | 4 -> Pair | _ -> Highest With jokers, it was slightly more options to match with. let toQuality2 hand = let jokerCount = getJokerCount hand let hitCount, maxEqualCardsCount = groupCards hand match hitCount with | 1 -> Five | 2 when jokerCount >= 1 -> Five //Five with joker | 2 -> //Four, House match maxEqualCardsCount with | 4 -> Four | _ -> House | 3 when jokerCount = 3 -> Four //Four with joker .... and more...

Great walk through. I tried using a masking technique in F# after seeing people in the APL community use that. It's worth a try, but I'm not familiar with the technique enough, so it was really painful. I saw some solutions in Python that used masking libraries as well and their solutions were only 20 lines and extremely elegant.

Todays puzzle was a breath of fresh air! Nice tip about enriching the types when piping 👍

Very cool idea of using the range. I actually was running into a tiny bit of a problem because initially I tried to fit everything into a map. while it worked for the test input, real data proved my decision very wrong xD thanks for the video!