Talk:Straddling checkerboard

The Wikipedia page implies that the order of the digits 0-9 can also be scrambled to improve security. To make the task more complete but not much more complicated, I think it would make sense to also provide numbers to the function in the desired order. To leave the existing cipher for the example valid, the inputs to the StraddlingCheckerboard function would become ("0123456789HOLMESRTABCDFGIJKNPQUVWXYZ./", 3, 7) --Tikkanz 15:05, 6 June 2011 (UTC)

You mean the order of the columns? Changing the order of the columns would be the same as rearranging each of the rows. Since you can input the whole alphabet in any order (as opposed to just a keyword to get things started, for example), it wouldn't add anything new. MagiMaster 19:44, 6 June 2011 (UTC)

It might be nice just to standardize the examples. Then we can have example input and output in the description for testing purposes/proof. --Mwn3d 19:58, 6 June 2011 (UTC)

I was having trouble finding some existing piece of text that was relatively short and had a few numbers and full stops. MagiMaster 20:49, 6 June 2011 (UTC)

Category

I hit the wrong button while I typing in the reason for the undo. Anyway, the category will be included automatically once this is promoted to full task status. MagiMaster 19:51, 6 June 2011 (UTC)

Stop character

I kind of like how it looks when using the stop character to (instead) represent any unencrypted characters (like the Util's Perl6 solution did). Should I put this as part of the specification, or leave it as is? Since this is supposed to be a demonstration of a historical method, how it was actually used should weigh in the decision, but I don't really know much about that. MagiMaster 04:07, 7 June 2011 (UTC)

From The SECOM Cipher (which uses a straddling checkerboard):

In our example, several spaces are used. However, it is recommended to omit all spaces where legibility preserved.

AFAIK, when you use *any* field cypher in real life, you make a first pass over the plaintext to reduce whitespace, using human judgement on where space must be maintained to preserve the meaning of the message. This improves the cyphertext's resistance to cracking. The cipher procedure itself is mechanical, and cannot presume to know what will be "legible" if words were to be run together. Space reduction is not helpful in examples, and not appropriate for this task.

The only reason I added the $collapse version in my Perl 6 solution was to show encoded&decoded output *identical* to the only other solution. I would prefer to remove the $collapse code.

I recommend these changes to the task:

• Specify the plaintext and key+u+v in the task.

• This will make it easier to check for correctness in the encoded output across all solutions.

• Choose the task's plaintext to contain a mix of only spaces, numbers, and uppercase letters.

• Perhaps: RV TOMORROW AT 1400PM TO COMPLETE TRANSACTION USE DEADDROP AS USUAL

• Choose the task's key such that the first row contains ESTONIAR in some order, and place . and / somewhere in row 2|3 that is *not* in the last two positions.
• Continue to assume an order of 0123456789, since the option of its reordering is redundant to reordering the alphabet in the key.
• Choose u and v where u>v, to catch solutions that miss (as the C++ solution does) the distinction in where to use original-(u,v) versus sorted-(u,v),
• Specify that space becomes stop char when encoding.
• Leave as unspecified any details on how to handle invalid chars in the plaintext. Let solutions die on invalid input, or translate to stop char, or whatever.

• Alternately, specify that non-alphanumeric chars be treated as whitespace, since that is what would probably happen if the field encoder was not the plaintext author.
• This point is moot if you limit the task's plaintext to only spaces, numbers, and uppercase letters.

• Allow for solutions to (when decoding) either convert stop char to space, or leave as stop char.

--Util 15:50, 9 June 2011 (UTC)

Min/max

I don't think min and max are in <algorithm>, at least not in Visual Studio. Is it different with GCC? MagiMaster 11:06, 7 June 2011 (UTC)

When I comment out the #include line for algorithm, and compile with GCC's `g++` command, I see no change in behavior. --Util 15:32, 9 June 2011 (UTC)

Rename?

Might this page be better named as "Straddling checkerboard encryption"? (Rationale: straddling checkerboard may have other meanings and interpretations in isolation). --Markhobley 11:40, 8 June 2011 (UTC)

We can jump that hurdle when we come to it. No sense in worrying too much about what only might happen in the future. –Donal Fellows 14:29, 8 June 2011 (UTC)

Ok, I had to examine the task to examine the nature of it contents. The encryption suffix would have saved a few moments. Is there any reason not to rename here? It can be done at the click of a button. --Markhobley 16:15, 8 June 2011 (UTC)

Shorter names are nicer/cleaner in lists and links. Proper categorization/tagging should avoid that confusion. Really, what else could it be? "Straddling checkerboard cipher" is probably a better name than the "encryption" one, but that seems to be an unnecessary specification like calling Linked list "Linked list data structure". --Mwn3d 16:27, 8 June 2011 (UTC)

Bug

Note that in the C++ solution, changing 3,7 to 7,3 results in identical encoding, indicating that rowU,rowV are being used in places that need u,v instead. --Util 15:32, 9 June 2011 (UTC)