Rijndael

[tormento]

I'm using Kremlin Text and I find its encryption features useful.

I'd like to see a secure and fast algorithm, such as Rijndael, implemented in Atlantis. A compression feature should be useful too.

Would you consider using it? (It's free..)

[admin]

Atlantis already has the encryption feature.
Click here to read about this feature of Atlantis:
http://www.atlantiswordprocessor.com/en ... uments.htm

[tormento]

I know it has encryption but we don't know anything about the type of and the security. I am not a NSA expert, so I'd like to know more about the encryption used.

The 2nd question (probably OT) was: what about a compression of the saved files? It should be very efficient, since it's above all a pure text file.

Thanks!

[admin]

Atlantis uses the PC1 algorithm (a stream cipher encryption with output retroaction).

>what about a compression of the saved files

You can save the desired document as a ZIP file (File | Save Special | Save As Zip…)
As you probably know, Atlantis does not have its own document format (except the encrypted document format – COD). Atlantis uses the standard document formats supported by other word processors. This allows to achieve maximal portability of documents created by Atlantis.

[maxqnz]

I know it has encryption but we don't know anything about the type of and the security. I am not a NSA expert, so I'd like to know more about the encryption used.

Thanks!

FWIW, I posted the algorithm details to a specialist cryptography newsgroup, and asked for opinions on its merits. Here was the response:
As far as I can tell, it isn't too bad - the overall scheme is very
similar indeed to that used by EFS (random session key used to
stream-encrypt a file, then a header prepended with that session key
encryped to a user key) with the modification that the chosen algo for
the session key encapsulation is Symmetric, not PKI, and moreover is the
same as used to protect the file.
The two algos in use (PC1 and PSCHF) were designed by Alexander Pukall
in the late 80's, and while I can't see anywhere a 256 bit "upgrade" of
the algo (he moved on with another algo with 160-2048 bit keys called
caracachs - leaving PC1 with 80 or 128 bit keysizes) it is presumably
just an increase in the same manner as 80-->128.
I don't see any attacks on the algo on the web (they may not be *on* the
web of course, it is a fairly old algo) and note that XChat have also
chosen PC1 as their stream algo.

[admin]

Alexandre Pukall is the original designer of the PC1 encryption algorythm indeed,
and also the Atlantis encryption expert. The Atlantis encryption feature itself
was implemented under the tight supervision of Alexandre Pukall.

Here are Alexandre Pukall's comments on your posting:

It is not true that PC1 can use only 128 bits keys. For example, Atlantis uses
256 bits keys. Atlantis uses a pair of two PC1 ciphers simultaneously
(synchronously) with different keys. But the PC1 cipher could also be modified to
use 256 bits keys with no problem at all.

Folks may say that it is a "fairly old algo", but no one has yet been able to crack PC1,
even though hackers had plenty of time to do so!

People who know nothing of cryptography always expect trendy names
that they hear through the Press or on the Web like Rindjael, Microsoft, or Coca-Cola.

But all through the years PC1 has proved to be as safe as other 256 bits algos.

[maxqnz]

Here are Alexandre Pukall's comments on your posting:

Thanks for that clarification. BTW, that was my posting, but not my words. I was only passing on what someone else said, as I am only too ready to admit that know next to nothing about cryptography. It was very reassuring to hear from Alexandre himself, as we say in English, straight from the horse's mouth. :lol:

[maxqnz]

Atlantis uses a pair of two PC1 ciphers simultaneously
(synchronously) with different keys. But the PC1 cipher could also be modified to use 256 bits keys with no problem at all.

Are there any plans to so modify PC1. If using two 128-bit keys simultaneously gives Atlantis an effective strength of 256-bits, then doing the same with two 256-bit keys would effectively give us 512-bit security, is that right?

[admin]

Have you ever heard that somebody cracked a 128-bit or 256-bit encryption?

[maxqnz]

Have you ever heard that somebody cracked a 128-bit or 256-bit encryption?

:)

I was just curious, that's all. Moore's Law suggests that what may not be feasible now maybe in the not-too-distant future.