OMGeeky/hcsalmon1-Chess-Engine-Test
1
0
Fork 0
mirror of https://github.com/OMGeeky/hcsalmon1-Chess-Engine-Test.git synced 2025-12-26 16:27:23 +01:00
Code Issues Packages Projects Releases Wiki Activity

Update explanation.txt

Browse Source
This commit is contained in:
Coding with Tom
2025-01-19 14:50:42 +00:00
committed by GitHub
parent 407e88cbe4
commit fd49de97b6
1 changed files with 4 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
Java/explanation.txt
View File

@@ -1,16 +1,6 @@
I had to change the algorithm with java. Instead of the standard rook and bishop move generation, I split
them up into direction and then get each direction individually. This is slower but stops me needing a 54x4096 and
64x512 array. This is a problem because ulongs are not directly supported in java.
them up into directions and then get each direction individually. This is slower but stops me needing a 64x4096 and
64x512 array. This exceeds the max size for a class, so I went with this approach.
You have two choices
1. convert all the constants to minus, if they are over the max signed long value, and risk a lot of bugs and testing that way.
2. convert all the constants to BigIntegers like this:
static final BigInteger MAX_ULONG = new BigInteger("18446744073709551615");
This means replacing all values with this type in all constants. You also need to use OOP for everything.
This:
BigInteger bitboard = bitboard1 & bitboard2;
Turns into:
BigInteger bitboard = bitboard1.and(bitboard2);
I chose the second approach and tried converting everything over to this. I created so many bugs that I couldn't fix and
gave up in the end.
I finally got java to work after loads of trial and error. Having to use signed longs can cause many problems in certain situations.
Like in bitScanForward.