I took a peak at some files. It looks like TAKP source is based on older EQEmu and has not been kept in a merged state. Specifically I wanted to know if they were using the min_expansion/max_expansion fields noted here:
https://docs.eqemu.io/developer/proj...-content-flags
These fields don't exist in the database so TAKP version predates this being implemented. Not sure if TAKP has some other mechanism they implemented to be able to roll content out for each era. I'm suspecting they don't and just implemented things as they occurred. For example POP released and all the code just switched to a new version. They might have older source versions tagged to eras but that wasn't released so this is probably the next necessary thing.
Someone needs to now use all the information TAKP dumped but organize it into eras including changes over time. This is a lot easier than getting all the "classic" quests implemented in the first place so this dump is still very useful. Now you just have to know which era to put things and how to change them over time for each era.
Dolalin made an EQ emu archive search engine but I'm thinking this info should instead be used for training an AI model. This is a side project I want to start soon, I wanted to look into llama home brew anyway. My thinking is giving the corpus of classic EQ info to a language model would allow it to create a summary of changes over time. "Tell me how the sneak skill changed over time" and get a list of changes for this one skill with references to patch notes or wherever the info may have come from. I've been reading others who are training language models on specific knowledge domains like this with some success.
This is probably never going to work but I'll know a little more and I'll be able to say I tried to create the ultimate EQ nerd AI. It would be funny to get it posting on the P99 forums.
07-31-2023, 09:49 PM
Hybrid Mode
