About hash string header bytes

[lyrachord]

Hi, the hash has 6 half-byte chars: checksum, Lvalue and Quartile, 3 bytes.
I have 2 questions:
1, why checksum is a byte array? it's just a single byte.
2, why do swapping for the head bytes? historical reasons?

Regargs

[lyrachord]

in tlsh_impl.h, more versions of TLSH_CHECKSUM_LEN found, so Q1:) is solved

[jonjoliver]

Could you explain what you mean by "swapping for the head bytes" ?

Cheers
jono

[lyrachord]

Thanks jono

swapping for the head bytes

I mean, swap_byte invocation in hash function

f.e. a digest with checksum[1], Lvalue, Q, which head hexes are 3 bytes:

C1C2 L1L2 Q1Q2

function hash() will return a string which starts with

C2C1L2L1Q2Q1.

there is a swapping-first, why?

Regards

[jonjoliver]

I think the swap byte invocation is required because different OS store the bytes in a different order
And the different calls allow the programs to get access to the header in a consistent logical view independent of the OS implementation of memory

[lyrachord]

Thanks.
It seems it's a old history. You know, new CPUs care no a single byte, just byte order, even more MCUs.
I guess, to replace the swapping way with current normal byte hex string representation and to keep the old hash in history function is not bad idea?

Regards

[jonjoliver]

Hi lyrachord
I don't think I will change the header code - I think it works in some older implementations
And as far as I am aware - the code as it is now - is consistent across OS - and is not breaking.
Cheers
jono

[lyrachord]

Hi jono
Breaking or not is a problem. You're right.
Let's in other words, Keeping the swapping and add a new version without swapping hash2 function?

BTW, you know the TLSH-java project deleted, I'm planning to contribute another java version. But there is naming convention inconsistent problem between java and c/c++, your advice?

Regards

[mrpolyonymous]

Hi @lyrachord
I just contributed a Java port to the TLSH project here, it's in https://github.com/trendmicro/tlsh/tree/master/java

It will look very familiar if you've looked at the TLSH C++ code, and for good or bad I resisted making the variable and method names more Java-like to keep the similarity to the C++ code.

The library isn't yet on Maven central or any other JAR hosting site, that's my next task. If you're okay building locally then running gradlew build will make the JAR file for you.