As long as it takes. At this point in time I see on Github a partial fixed point compiler-compiler and an Earley parser.
See the Earley parser: https://github.com/IDNI/tau/blob/master/parser/earley.cpp
From what I can see the Earley parser is no where near ready or complete. It's in the draft stage where it's being worked on. Earley parsers deal with formal grammers, and specifically context free grammars, which are essential for formal languages such as programming languages. An Earley Parser simulates a non-deterministic finite automaton essentially.
So we have to look at the current state of Github: https://github.com/IDNI/tau
And the promising aspect is that the design for Tau appears complete. The implementation phase occurs right now. I cannot know how quickly or slowly it will take Ohad to write the code but I do have a rough idea of what kind of code needs to be written. When you consider testing and so on, we could see the first alpha this year at some point. I don't think we will see the beta this year.
It took two years to write something that a student could code in two hours days. IMO Tauchain is a scam.
What have you ever coded? Do you really think myself and all these others who have coding experience and degrees would hang around Ohad for 2 years if he wasn't onto something? If you think what he is doing is so easy then I challenge you to go beat him to it. Someone please go build an equivalent to TML and I will jump all over it.
You have to understand that the genius of Ohad Asor is not so much in his ability to code. Anyone can code once the design and specification is complete. The genius is in the 2 years of R&D that came up with that design! Just hacking together a parser? You're right, even I could do it as a hack. That doesn't mean I would get the design right which is where the actual challenge is.
For instance with Early Parsers if you don't understand why these particular parsers parse in cubic time (worst case) which means if the input is twice as long then parsing takes eight times as long, or that the benefit of these parsers is that it can parse ANY language, then you will not even understand what TML is trying to do. To say anyone could do it? It indicates to me that you don't actually understand what TML really is.