Puzzle comparisons

19 posts / 0 new
Last post
ben
Offline
Joined: 2012-04-02

I had been thinking that something like this was possible...
It would have saved an entry and still given the same data.

Yep, Sokodup can have better formatted output.
If it does a second pass, or structured the stats in a different way, it's possible to show better output.

We are talking about puzzles that are known to have a more unique type of solution, and one for which duplicating it in an altogether different puzzle would be more work than anyone would want to put into to designing such a puzzle. In addition, we are talking about puzzles where the similar puzzle is so close to being the same ...

Again, I was just saying you cannot claim levels are the SAME using solution like that. They are similar, yes, no question about it.

I made these this way for visual comparison. You can easily tell where the differences are, and much more quickly, seeing them together than when having to go back and forth between puzzles to examine them....

I lay them out side by side, I cannot tell that they are the same.
It's easy to tell that they are similar, but no way to tell if they are the same until you look at every single unit composition of the level.

If you recall about the Boxxle hidden level, you will know.
I can immediately tell that it's an easier version, no mistake.
and you didn't find any difference in them at first.
That's because looking SIDE by SIDE is hard to tell.

If you say that toggling between two images is not easy to tell...
that's because you haven't really seen it in action yet.
I have tried both and I can tell you that graphical comparison toggling back and forth within the same window is very easy to tell what's changed. Viewing side-by-side is not easy -- unless you have a DIFF program that shows you what's different.

It is like you view two levels one no-top of another, it's very very easy to see what's changed.
I wish this site allows attaching animated GIF so I can show you.
If you provide an email, I will send it to you.

WayneCa
Offline
Joined: 2014-07-13

I went through the Duplicates.sok file and made some better notes. If you use YASC you can see the notes for each puzzle by clicking the Open button and looking in the info pane as you scroll through the puzzle titles. The main file notes can be read in a text editor. I notice that YSokoban does not seem to have a function for viewing level notes, and therefore I assume that feature isn't supported. In the case that whatever engine you use to play sokoban does not support the level notes feature, you will have to look at them in a text editor.

For what it's worth, I use TextPad. It is a much better editor than Notepad and allows me to use a mono-space font that makes looking at the puzzles much better. TextPad was developed as a programmer's code editor that works with plain text files, so it is perfect for use as a sokoban level file editor. I recommend it highly.

I am attaching the updated Duplicates.sok file here (with the .txt extension).

attachment: 
WayneCa
Offline
Joined: 2014-07-13

If you say that toggling between two images is not easy to tell...
that's because you haven't really seen it in action yet.

I have tried both and I can tell you that graphical comparison toggling back and forth within the same window is very easy to tell what's changed. Viewing side-by-side is not easy -- unless you have a DIFF program that shows you what's different.

I have done this before. I originally made the tribute file so I could view the similar puzzles by having them adjacent to each other and switching back and forth between them to see the differences. It works, but it is more work than I want to do (and takes more time than I want to spend) when I can look at them side-by-side and find the differences almost immediately. I used the toggle method for Boxxle 1 hidden and perfect 303 by having 2 instances of Sokoban YASC open with each at the desired puzzle and switching back and forth. I almost gave up and was about to start creating a side-by-side view. Then I thought, why not put both windows side-by-side. That's when I spotted the difference.

For me the side-by-side view is better and easier. I guess it's one of those things where what works for one person does not work for another.

ben
Offline
Joined: 2012-04-02

I used the toggle method for Boxxle 1 hidden and perfect 303 by having 2 instances of Sokoban YASC open with each at the desired puzzle and switching back and forth

That's exactly what I said when I say you haven't seen it in action yet.
You do not open two instances or two windows.
If you have to open two windows and then compare like that, that's no difference than looking at them side-by-side.

You toggle two levels within the same windows.
For example, if you use YASC you toggle the two levels within one YASC instance.
I use YSokoban. but I believe the same thing can be done in YASC.

WayneCa
Offline
Joined: 2014-07-13

After I uploaded the file I realized I had made an error in the level notes for level 42. The line that says "Commodore 64 repositions them at column D, row 9 and column G, row 11, creating a deadlock" should read "column F,row 11".

WayneCa
Offline
Joined: 2014-07-13

I know what you meant. I have created files with the necessary puzzles adjacent within the same file and just kept clicking the next and previous arrow buttons to flip back and forth between them. It does work, as I said, but it's more effort than I want to spend. For me, if I have to create a file containing both puzzles anyway, having them in one puzzle that I can study at the same time, with no flipping, works better.

ben
Offline
Joined: 2012-04-02

For YSokoban, there's no need to create one file to contain all levels you want to toggle between. You can toggle between different levels in different sets. I will drop this topic here.

I know when it comes to using programs and tools, it's hard to convince people. It's best to use whatever you're comfortable with.

WayneCa
Offline
Joined: 2014-07-13

I spent some time this morning going over the file notes to the Duplicates.sok file. I noticed that there were many things that needed to be changed and other things that needed to be added. Instead of uploading the file here again, I am going to put the file on a blog, and here I will simply post the new version of the file notes.

As an example to back up my assertions, look at the partial puzzle below. Which one did I copy, Boxxle 1 hidden or Perfect 303? Note that I made the area that is different between them all spaces so you can't tell by looking which one it is.


  

::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: There are 34 puzzles that are duplicates between :: :: the original puzzles (all 4 variants + xsokoban), :: :: Boxxle 1 and 2 and Perfect. There were no duplicates :: :: found in Revenge. :: :: :: :: In order to avoid a problem with the start position :: :: character being removed in a puzzle if more than :: :: one exists, I have replaced all start positions :: :: with boxes on goals. Only two puzzles have boxes on :: :: goals from their start configuration, and in both :: :: cases those occur in the goal area. In all cases, :: :: the start position is outside the goal area. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: The Boxxle 1 and 2 collections currently found on the :: :: Internet in text file form display the puzzles they :: :: contain as vertically flipped when compared to the :: :: Originals and the GameBoy version of those :: :: collections. :: :: :: :: Sokoban Online displays the Boxxle 1 and 2 levels :: :: as in the same orientation as the originals except :: :: as stated below, and ben from the Sokoban for :: :: Windows forum has compared the Boxxle 1 and 2 :: :: collections to his Japanese and US rom carts for :: :: the GameBoy version and has stated the same result. :: :: :: :: I have ignored vertical flip re-orientation when it :: :: is the difference in a Boxxle 1 or 2 puzzle. Where :: :: re-orientation is different otherwise, the :: :: re-orientation is stated for the puzzle(s) it :: :: applies to. :: :: :: :: In describing wall block, box and goal differences :: :: I have used capital letters to denote column :: :: position and numbers to denote row position. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: Note that YASC Duplicate finder and SokoDup both :: :: missed the exact same duplicates. This is due to :: :: the currently accepted definition of what :: :: constitutes a duplicate and not a fault on the part :: :: of the author or the program. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: Boxxle 1 Level 81 is rotated 90° counter-clockwise. :: :: Note that since I used the vertically flipped :: :: version in this document, the puzzle seems to be :: :: flipped, but in the GameBoy game it is not. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: Boxxle 2 Level 33 is rotated 90° clockwise. Because :: :: of the nature of this puzzle, the flip makes no :: :: difference to the orientation of the puzzle. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: Perfect Level 71 is rotated 90° clockwise. :: :: Perfect Level 77 is rotated 90° clockwise. :: :: Perfect Level 146 is rotated 90° clockwise. :: :: Perfect Level 163 is rotated 180°. :: :: Perfect Level 241 is rotated 90° clockwise. :: :: :: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: :: :: A different view of duplication: :: :: :: :: There are certain puzzles that are duplicates save :: :: for slight modifications which will render a :: :: different solution. In those cases where the :: :: modification is the addition or removel of one or :: :: more boxes/goals, the puzzle is considered unique :: :: as the solution is unique and will not solve the :: :: unmodified variant, nor will the solution to the :: :: unmodified variant solve the modified puzzle. In :: :: the remaining cases where the modification is the :: :: addition or removal of a wall block, more thought :: :: is required. The solution to the less difficult :: :: puzzle can be significantly different as to make :: :: that solution not applicable to the more difficult :: :: variant. However, the solution to the more :: :: difficult puzzle is applicable to the less :: :: difficult puzzle and will solve that puzzle. :: :: Therefore these should be considered as duplicates. :: :: Note that this definition of duplicate is based on :: :: the puzzles being identical in every other way :: :: beyond the minute differences. I leave it to others :: :: to argue the merits of this point of view. :: :: :: :: My strongest arguments as to the idea that these :: :: puzzles are duplicates are: :: :: :: :: 1 If you don't know what the differences are, :: :: they appear to be the same puzzles to the human :: :: eye. It is humans that the game was designed :: :: for, not computers, and therefore the :: :: appearance to the uneducated human should be :: :: the determining factor. :: :: :: :: 2 All of the puzzles began as single puzzles :: :: which were then manipulated by the software :: :: companies. This is where the term "remodel" :: :: comes from as it applies to sokoban puzzles. In :: :: the case of the originals, the modified :: :: versions occupy the same position in the :: :: collections they occur in as the first original :: :: does, and are known by the same level number as :: :: the original. In doing so, they become the :: :: replacement original, and are the same puzzle, :: :: only remodeled. :: :: :: :: If, then, either the original or the remodeled :: :: original are included in a different collection,:: :: whether further remodeled or not, they :: :: constitute duplicates of the original. :: :: :: :: This definition of duplication is not widely :: :: accepted. Due to the currently accepted definition, :: :: most people asked will disagree with this :: :: definition. My position is that as time changes, so :: :: do definitions, and at some point this definition :: :: will become the "currently accepted definition". :: :: :: :::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Pages