We also need better names for this.
If you take the current set: hybrid2(alpha#1-1,loweralpha#5-5,loweralpha-numeric#2-2,numeric#1-3) then encoding it in RTI2 as Sc00bz intended (and what is a hybrid3 construct): hybrid3(hybrid2(alpha#1-1,loweralpha#5-5,loweralpha-numeric#2-2,numeric#1-1),hybrid2(alpha#1-1,loweralpha#5-5,loweralpha-numeric#2-2,numeric#2-2),hybrid2(alpha#1-1,loweralpha#5-5,loweralpha-numeric#2-2,numeric#3-3))
This is 3 sub key spaces composed of each hybrid2 that contains sub sub key spaces. Now, to be fair something like an Omni set won't ever fit in 255 characters of a file name and so hybrid3 is likely to be an internal name of sorts that isn't exposed to the user.
Within the RTI2 naming each sub key space is composed of 1...N hybrid character sets which refers to a character set at a given length. So encoding one of the hybrid2s is three sub key spaces with 4 hybrid character sets of lengths 1, 5, 2, and 1 (2 and 3 for the 2nd and 3rd sub key space).
In a sense you can think of hybrid2 as what hybrid was intended to be in the first place. Then, thinking of hybrid3 as the full RTI2 sub key space implementation or as the enclosing/"super set" construction. Of course making a clear RTI2.0 specification is difficult enough and all of this will eventually become RTI2.1. To have to describe hybrid3's that enclose hybrid2s of fixed length and that hybrid3 has sub key spaces, hybrid2 has sub sub key spaces, and that each hybrid 2 sub sub key space actually refers to RTI2 hybrid character sets...just naming even if internal to the construct, format, and spec need to be understood by more than just Sc00bz and I :D
It's ok to nod and smile and just ignore this post since Sc00bz and I spent some time rather perplexed with each other and finally had a real time conversation in which we both realized we weren't talking about the same things.
I'm going to attach a draft copy of the RTI2.0 spec WARNING DRAFT
and in some cases I just threw stuff on the end. This spec actually has what is hybrid3 at the end *and* the character/sub key space encoding on disk for the RTI 2.0 releases so far has incorrect header encoding and the file name must be used for those bits. This copy is only being attached for review and comment and *should not* be used for implementation of the format. This attached specification should be considered public domain.Update:
removed the attachment as the spec is now on the download page or directly via this link