# .font: font.text_2
2024-03-11 21:35 dustin  
# .font: font.text_1
very thoughtful and challenging cave again! :d  
  
as my run (for example) shows, the second Ζ crushing can be avoided, so the corresponding race against the ΜΜΜ can be skipped i guess that's unintended?!

# .font: font.text_2
2024-03-12 09:42 heywood  
# .font: font.text_1
thank you! IT was unintended, i triggered that a few times during testing but couldn't work out how to modify the cave to prevent it. i guess it's always possible for the expanding wall to sneak in behind you...

# .font: font.text_2
2024-03-12 20:59 dustin  
# .font: font.text_1
THIS IS ACTUALLY A QUESTION OF THE CAVE SCANNING ORDER (THE TOPIC OF MY "CSO" SERIES), SO IT'S A PLEASURE FOR ME TO GIVE YOU SOME INFO :D  
  
IN CASE OF Κ BEING FOLLOWED BY A GWALL, THE RULE IS "RIGHT BEFORE BOTTOM BEFORE TOP BEFORE LEFT". THIS MEANS:  
  
| IF THE GWALL IS TO Κ'S RIGHT SIDE, IT WILL ALWAYS SNEAK BEHIND HIM, THERE'S NOTHING THAT CAN STOP IT.  
  
| IF THE GWALL IS AT Κ'S BOTTOM, IT CAN ONLY BE STOPPED BY ANOTHER GWALL TO Κ'S RIGHT. FOR EXAMPLE:  
  
ΕΓΕΕ  
ΕΓΕΕ  
ΕΓΕΕ  
ΕΓΕΕ  
ΕΓΜΕ  
ΕΓΕΕ  
ΕΓΕΕ  
ΕΚΕΕ  
ΕVΕΕ  
ΕΕΕΕ  
(THE "V" IS A VERTICAL OR BIDIRECTIONAL GWALL.) NOW IF Κ MOVES UP, THE HORIZONTAL GWALL WILL STOP THE VERTICAL ONE BECAUSE "RIGHT BEFORE BOTTOM".  
  
IF YOU FLIP THE ABOVE "MINI MAP" HORIZONTALLY, THEN THE VERTICAL WALL WOULD NOT BE STOPPED BECAUSE "BOTTOM BEFORE LEFT".  
  
| SO IN CASE OF THE SCENARIO IN YOUR CAVE...  
ΕΕΕΕΕΕΕ  
ΕΕΕΒΕΕΕ  
ΓΓΓΓΚΜΕ  
ΕΕΕΙvΕΕ  
ΕΕΕ ΕΕΕ  
...the bidirectional wall "defeats" the v-wall because "right before bottom". if the whole cave was horizontally flipped, then "bottom before left", the v-wall wins and "cheating" would not be possible.  
  
ALSO NOTE THAT IF THE GWALL IS TOP OR LEFT FROM Κ, THEN IF Κ MOVES, THE WALL WILL NOT IMMEDIATELY FOLLOW, SO THERE WILL BE A SMALL GAP BETWEEN Κ AND THE WALL. so one more scenario to demonstrate this:  
  
ΕΕΕΕΕΕΕ  
ΕΕΒΕΕbΕ  
ΓΓΓΓΓΚΕ  
ΕΕΕΕΕΕΕ  
(b=bidirectional gwall)  
here if Κ moves left, the Β will block the gwall despite "right before top", because at the beginning, the b-wall cannot immediately follow Κ but only with a 1-space gap. however, if Κ hesitates after the first step, letting the b-wall catch up, then "right before top" is again valid so the b-wall won't be stopped by the Β.

# .font: font.text_2
2024-03-12 23:22 krissz  
# .font: font.text_1
 dustin , we have Ρ character!

# .font: font.text_2
2024-03-13 01:26 heywood  
# .font: font.text_1
 dustin , this is very informative, thanks for taking the time to write it.   
  
something i'd really love to learn is how to consistently exploit the scanning order for pushing regular boulders across gaps. see the comments / demos on "no one's boulder 23 cave O" for context. i spent a huge amount of attempts trying to get consistent timing by watching the butterfly, but the random push time before boulder actually moves seems like a wild card.  darkstoorm 's demo is aspirational.  
  
thanks again.

# .font: font.text_2
2024-03-13 11:36 dustin  
# .font: font.text_1
 krissz : i know, i tried to make it but it didn't work (although it does work in the ckit). maybe some technical issue?! i also have sometimes problems with posting links in a comment because the letters appear uppercased. there must be some issue between my laptop/browser and the script used for this site :d  
  
 heywood : my pleasure! in case of pushing a Β over a gap, the cave scanning order determines that it is possible only from left to right, not vice versa. and even then, as a Β can be pushed only in 1 out of 4 cases, the chances that it works are 1/4 = 25% for each gap. that means for the no1 23-o cave, where Κ tries to push the Β over 3 gaps, the chances are (1/4)*(1/4)*(1/4) = 1/64, this is practically "true" randomness, so there's no way to determine if it will work next time or not, the only way is to keep trying until it works (while in addition, the pushes also have to be timed perfectly so that the Β will hit the Η afterwards!) so that's why that cave has so many attempts (i assume that even without your attempts, it would still be many!) ;)   
  
note that  krissz  has added light ΠΠΠ in the ckit mainly for the purpose to make these over-gap-pushes deterministic, which makes caves possible where a Π must be pushed over the whole cave without requiring ten million attempts ;)