![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
Posted here due to the amount of discussion going on re: whether a version of Rybka has code copied from other programs, such as, Crafty, Fruit, etc.
I believe McCabe is well known within the software industry as a supplier of a decent set of tools to allow the examination of code; I believe one of these is specifically aimed at 'comparing' code. Perhaps someone who posts on this forum has experience of these tools and can comment from a practical experience rather than what is posted on the McCabe website. {http://www.mccabe.com/}
Perhaps the ICGA (and other organisations wanting to enforce a 'NO COPY' regime on engine participants) might also want to take a look at this type of solution to the problem as it has the potential to remove any subjectivity resulting from one set of experts claiming A = B and another set claiming A ≠ B
I believe McCabe is well known within the software industry as a supplier of a decent set of tools to allow the examination of code; I believe one of these is specifically aimed at 'comparing' code. Perhaps someone who posts on this forum has experience of these tools and can comment from a practical experience rather than what is posted on the McCabe website. {http://www.mccabe.com/}
Perhaps the ICGA (and other organisations wanting to enforce a 'NO COPY' regime on engine participants) might also want to take a look at this type of solution to the problem as it has the potential to remove any subjectivity resulting from one set of experts claiming A = B and another set claiming A ≠ B
![[de]](/mwf/v12/flags/de.png "Germany")
It looks to me like this tool works on source code level and it's only designed to find redundant code within one program. So it can't compare 2 executables.
OK, what about bitmatch here {http://www.safe-corp.biz/products_codesuite-lt.htm} - any good?
This is from their website:
So this also won't work.
>BitMatch examines all text strings, comments, and identifier names that it can find in the executable files in order to determine copying. If a specific user message or a unique subroutine name is found in two files, there is a possibility that one was copied from the other. Note that BitMatch gives only a rough determination whether copying took place. False positives and false negatives are both possible. CodeMatch is needed to compare source code to make a definitive determination.
So this also won't work.
![[us]](/mwf/v12/flags/us.png "United States")
Hi Lukas,
could you tell us anything about how Vas is doing these days and what he is up to? Thank you.
could you tell us anything about how Vas is doing these days and what he is up to? Thank you.
It helps. I wrote my own "hex string search" code to locate what I consider to be "unique Crafty constants" in the various early Rybka binaries.
But that kind of matching only goes so far. One still needs to disassemble to expose the semantics. Finding the strings is something that can be used to tell you whether the disassembly process is worth considering...
There are really good free tools available to compare semantics between two different sources. Several universities have written them and there are quite a few papers on the topic since this is a continual problem in computer science and information science type departments... but if you don't have both sources, things get _much_ harder.
But that kind of matching only goes so far. One still needs to disassemble to expose the semantics. Finding the strings is something that can be used to tell you whether the disassembly process is worth considering...
There are really good free tools available to compare semantics between two different sources. Several universities have written them and there are quite a few papers on the topic since this is a continual problem in computer science and information science type departments... but if you don't have both sources, things get _much_ harder.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill