Where are the best computer chess experts?!

[Jouni]

I think most of them are here at talkchess. So why is it so difficult to clear if Robbo/Ippolite is clone or not Please hurry up and do it soon, so we don't waste time in testing them!

Jouni

[gerold]

I think most of them are here at talkchess. So why is it so difficult to clear if Robbo/Ippolite is clone or not Please hurry up and do it soon, so we don't waste time in testing them!

Jouni

The source code to Rybka has been lost. How can anyone be sure.

Other people have Fruit's code so someone will make a new
stronger engine than Robb.

[zullil]

The source code to Rybka has been lost.

This is unbelievable!

[Sven]

I think most of them are here at talkchess. So why is it so difficult to clear if Robbo/Ippolite is clone or not Please hurry up and do it soon, so we don't waste time in testing them!

Jouni

The source code to Rybka has been lost. How can anyone be sure.

I don't think VR has ever said that. I think he said he does not have the exact sources of the Rybka version anymore that was released as Rybka 3. But of course he has more recent source code versions.

To do a comparison of two sets of source files serving for the same overall task (chess engine) would still be non-trivial even if the exact sources on both sides would exist. It can be done by applying common indentation and other coding style rules to both sides (there may be good tools for that part), and then finding an "appropriate" mapping of identifiers for functions, variables, types and so on. However, the details are not that simple.

Even a harder task is to do the comparison based on one set of source files and one executable. The following might help in the specific case we are discussing here:

- having a copy of the Rybka 3 executable (for instance, I don't have one );
- knowing exactly the compiler version and settings that were used to build Rybka 3;
- having the earliest possible version of Ippolit sources (not Robbolito or anything newer since these were derived from Ippolit);
- having deep experience with decompiling tools like IDAPro;
- having decent knowledge about the features and properties of Rybka 3;
- finding an "appropriate" mapping between identifiers of functions and variables used in the sources on one side and addresses of functions and data in the executable on the other side, while being unable to fully derive things like data types, inline functions or #define macros from the executable;
- having enough time to investigate into such a huge task.

Sven