Markov stats questions
#1
First, thank you for implementing Markov mode - even the default oclhashcat-plus 0.14 implementation gets may more hits in the first part of the keyspace than in the last.

I have a few questions about the details of (the upcoming) Markov modes:
1) The current hcstat file appears to be of fixed size - is it reasonably practical to have an option for larger stat files?

2) My major current thoughts on Markov mode is that its primary new benefit is to run time-limited attacks on portions of the keyspace previously not worth pursuing - i.e. very large keyspace areas (full hex character set at "lower" sizes, and limited character sets at "larger" - 10 char and up).

3) How does the stats generator and stats processor work when "trained" on, and then required to output, say, a dictionary of very long words which only contain the ?d?s or ?d?l or whatever characters? What about a dictionary based on keywords found from a mask?

4) Per-position question: Does it differentiate "last" and "second to last" from 5 and 6 for 6 char long keyspaces, or 9 and 10 for 10 char long keyspaces?
#2
Please read this to get a first Idea of how to use it: https://hashcat.net/forum/thread-1291.html

1) No
2) There are different models. oclHashcat works with a different one, see thread above.
3) You can train it using hcstatgen in the hashcat-utils
4) Yes