https://kotlinlang.org logo
Join Slack
Powered by
Day 10, if you’re stuck - no solutions :christmas_...
# advent-of-code
p
Day 10, if you’re stuck - no solutions 🎄 While debugging my code, I made this test case, sharing it if someone else gets stuck too 🧵
K 3
Copy code
F-7..
L-J..
.S-7.
.|.|.
.L-J.
.....
This should lead to a 4 for part 1
kodee loving 3
Okay, spoiler for part 2 (rot 13 it if you want to know it 😁 )
Copy code
<uggcf://ra.jvxvcrqvn.bet/jvxv/Cbvag_va_cbyltba>
thank you color 3
👍 1
a
@Paul Woitaschek that's 4 for Part 2? How?? I probably missed something in requirements
l
Probably is 4 for Part 1
a
oh ok, that makes sense
it's just Part 1 is relatively straightforward .. Part 2 is apparently now
I was beginning to think about wrapping a map of a sphere or something 🙂
l
Wow, I mean looking at solution thread there is wide variety of possible solution.
a
but there is a solution, right? just checking because my seemingly correct code doesn't work 🙂
l
Ah yes but I mean people took different approach for the same result
I did the same as Paul. So you can look at his hint if you are stuck.
a
I didn't do anything fancy, just BFS
p
Yes the test case was just for part 1 on a stupid thing I was stuck with which was not covered by the test input
a
ok, makes sense now - i was like where "4" is coming from? 4-th dimension or somethin? .. )
p
What is your part 2 approach which isn't working?
a
i run BFS for points on the right side and on the left side of the path vector
if DFS hits the wall I know it's "outside"
p
Did you account for the case that you can squeeze in through the pipes?
👍 1
|| || || This shouldn't block you going up
a
it doesn't
i just walk the path, and whenever I can start a dsf (from left or right) I do it
test cases work
any unusual test examples you can share?
p
Construct a test case where there is a 3x3 free space inside
a
like this?
Copy code
F---7
|...|
S...|
|...|
L---J
p
Yes
a
9
it's 9, right? )
p
Yes ^^
a
yep, my algo works
p
Now try a bigger one where there is an inner field only accessible by two adjacent vertical pipes
d
this is the key test that needs to pass, from the description:
Copy code
..........
.S------7.
.|F----7|.
.||....||.
.||....||.
.|L-7F-J|.
.|..||..|.
.L--JL--J.
..........
the answer is 4
a
yeah, all the tests from description pass for me
@Paul Woitaschek you mean something like this?
Copy code
F---7
|...|
SF-7|
|L-J|
L---J
my code gives 9 because it calculates number of tiles inside big loop, regardless what's that is. Unless there is requirement somewhere not to count other inner loops, is there?
and the inner loop area is not connected with outside (nothing like || or = there)
i guess it's going to be my "the one that got away" 🙂 i'll try my luck with 11th day
d
My solution only considers pipes in the path
a
mine too.. the problem for me that there is no test case for me that fails except the big one
d
Why only left or right, not up or down?
(I used a diff algorithm)
a
what do you mean up or down?
d
I don't fully understand your algo
i just walk the path, and whenever I can start a dsf (from left or right) I do it
a
oh.. clockwise and counterclockwise - from the left side of the path vecor and from the right side
left and right relative to direction
of the vector
d
ah
a
and if DFS hits outside wall, it's outside of the loop. I know it's not the fastest and not mathy, but it works fast for the given input, just not the expected result 🙂
d
how does it squeeze thru parallel pipes
guessing something is wrong with that part
i.e. doesn’t work for all cases
a
it doesn't need to squeeze, it initiates the DFS alongside the whole loop
d
i don’t follow, how does it it reach an outside wall from the seemingly interior points here? DFS not only on open points presumably, how does it identify successor points?
Copy code
..........
.S------7.
.|F----7|.
.||....||.
.||....||.
.|L-7F-J|.
.|..||..|.
.L--JL--J.
..........
a
so there is a look, starting with S. imagine that there is right side and left-side DFS that begins at every point of the loop (every tile)
DFS stops only if it hits the loop itself or the wall
obviously accounting for already visited tiles etc
d
I still don’t follow, it seems like if you start at any O point here it will stop bc it hits the pipes and never hits the wall
Copy code
..........
.S------7.
.|F-*--7|.
.||OOOO||.
.||OOOO||.
.|L-7F-J|.
.|..||..|.
.L--JL--J.
..........
looking from the * point for example
a
Copy code
..........
.S------7.
.|F----7|.
.||OOOO||.
.||OOOO||.
.|L-7F-J|.
.|..||..|.
.L*-JL*-J.
..........
from * point, it will find internal area
d
yes that’s not the problem, the problem is the O points
you said your code passes but i don’t understand how
why doesn’t it flag them as inside
a
i don't need to worry about that - as long as there is a single escape DFS that hits the outside wall, it's considerd outside-side
d
what’s the DFS that hits the wall?
in this case
a
if you start with S moving right. your vector will be "->" on your left side there will be "outside" because there will be at least one branch of DFS that hits the outside wall
d
yes that doesn’t explain why it detects all the O points as outside
a
because O-points will be on the left side
so they are marked as outside
d
ahhh
I get it
you determine if L or R is outside
a
there could be only one outside
yep
d
then you fill on the opposite side
a
yes, as soon as I know that certain side is "outside" side, I just stop DFS there
and just count all the tiles on "inside" side
d
what do you get for this
Copy code
..........
.S-7......
.L7L7.....
.FJ.L7....
.L7F-J....
..LJ......
a
1 - is it correct?
d
yes
checking what would happen if an open point never touched a | or -
a
my part 2 just works with list of coordinates. with two points I have a vector that I can rotate left or right. the only Char that I'm checking is that it's not a wall tile
d
Are you counting interior points or are you counting exterior points and subtracting that and the pipe points from the total points?
a
counting interior tiles
d
I guess both ways would give the same result since something is being incorrectly flagged
My thought was, from a L type point there are 2 points on one side and 0 on the other
As opposed to one on each side
a
my assumptions: 1. there are points that on two different sides (left and right) 2. those sides could be either "outside" or "inside" 3. if there is even a single point that is the wall, that makes the side's of the point an "outside"
4. all "inside" points are reachable from the loop, e.g. for point inside point A, there is point L on the loop that could DFS into it (where loop line is the only blocker for DFS)
d
where’s your code
is your answer too high or too low?
(I assume too low)
a
i think it was just saying incorrect
d
well i can run it on mine I suppose
a
"That's not the right answer."
thank you!
I thought there might be some "off by one" error and I tried +/-1 as the answers and It said that "Your answer looks like someone elses puzzle answer" ) funny )
d
don’t write 
KFunction1<Pos, Pos>
, write 
(Pos) -> Pos
you’re 5 too low on my input
let me see if i can find which ones…
a
that's IDE"s func extract refactor - blame JetBrains folks ) but yeah, the code is not pretty
thanks - there must be some weird configuration of points/pipes
Copy code
countOnSide(
essentially returns a list of tile coordinates, you can see there what points are missing
d
yeah i found the points for my input
trying to print
a 21 by 21 region, missed point in dead center
Copy code
--JLJF7L-7F-7||LJ|||
--7F-JL-7LJ.LJ|F7LJ|
7.LJF---JF7.F7LJL-7|
L-7FJ.F--JL7|L7F--JL
-7|L--JF7F7LJFJL---7
-JL----JLJL--JF-7F-J
----7F7F7.F-7.|FJL-7
-7F-J|||L7L7|FJL7F-J
7|L--J||FJFJLJF-JL-7
J|F7F-J|L7L-7FJF7F7|
7LJLJF-JFJ.FJL7|||LJ
L7F-7L7FJF7L-7||||F-
FJL7L7|L7||F7||||||F
L7FJFJ|.||||||||||||
-JL7L7|FJ|LJ||||LJLJ
7F7|FJ|L7|F-J|LJF---
J|LJL7|FJ||F7L-7L-7F
FJ.F7LJL7|LJL7FJF-J|
JF-JL---JL--7||.L-7|
-JF7F7F7F7F7||L7F-J|
F7|||||||LJLJ|FJL7FJ
bordered by L J 7 F
let me repro simplest
Copy code
..........
.S-7......
.L7L-7....
.FJ.FJ....
.L-7J.....
..LJ......
that has L J F 7 in the same pos
relative to
the missed point
a
ok, let me try that
wait, is it a valid input at all?
d
oops wait
will fix
Copy code
..........
.S-7......
.L7L-7....
.FJ.FJ....
.L-7|.....
...LJ.....
that should work
now i realize i can run your code lol
it missed it
so, fix that
I’m guessing you aren’t looking out both “rights” at a turn
this point, you always approach head on and turn
that’s what I meant above, 
My thought was, from a L type point there are 2 points on one side and 0 on the other
a
fixed it ) thanks you! I needed two turning vectors )
🎉 1
d
yep
a
I should have tried that earlier when I saw that I get different result whn I attach turning vector to another point on a path vecotor
anyway, thank you so much for debugging it with your input and insights!
👍 1
d
nice algo
a
i'm sure there are some O(1) mathy ones - but I didn't want to open wikipedia on weekend )
d
I projected the path into a 2x size space and then used normal fill from (-1, -1) and then counted the even points that weren’t in the fill or path
19 Views
Open in Slack
PreviousNext
Powered by