summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2017-02-13Seems to work reasonably well.code-improvementsNathanael Sensfelder
2017-02-09It's starting to "properly" reply...Nathanael Sensfelder
The ACSL coverage is far behind though.
2017-02-03Continuing Implementation...Nathanael Sensfelder
2017-01-31Made Frama-C give up.Nathanael Sensfelder
It validates (*sequence_length == (\old(*sequence_length) + 1)) regardless of the addition of an "*sequence_length -= 1;" in the code.
2017-01-20Moving towards a server/clients structure.Nathanael Sensfelder
2017-01-18Done with 'core', starting to work on 'knowledge'.Nathanael Sensfelder
2017-01-07Starting to rewrite the knowledge stuff.Nathanael Sensfelder
2017-01-07Starts near-complete rewrite for code improvement.Nathanael Sensfelder
2016-09-21Fixes "No back sequence" + forbids name as pillar.Nathanael Sensfelder
2016-07-24Experimenting with K Order Markovian chains.Nathanael Sensfelder
I do not recommend using this branch ATM, it has not been tested.
2016-05-05Adds the current code.Nathanael Sensfelder
It's been running for close to a month on one of the IRC channels I frequent and seems to be working fine. One should be aware that, among other missing features, this version does not store permanently what the bot learns. Indeed, I am currently using a file with 431848 lines as its initial knowledge bank, making this particular feature not a high priority one. Also consider the fact that Zero of One converts text to underscore before reading it but will not change its own aliases. This could potentially be a cause for surprises when using uppercase letters in the latter.