Organizing Melee Fix (Programmers Read)

[Tollen]

Devs have offered to send the source code for this in the patch thread. I too don't have any programing background but my brother does and he used to play EQ in the 1999's with me so Ill give him a call and see if he would be up for it.

C++ I know he does, not sure about perl

[Bahleted]

I know my post was brief and not very explainative it was mearly an intro to getting something started.

I'm asking for help from people who do have the knowledge and time to help fix it or at least provide a productive medium in where this can be accomplished. I fully understand people might not have the time to contribute I'm not expecting anything from anyone who doesn't have that time.

Aeolwind pretty much implied he would be willing to share code with people willing to try due to lack of manpower. Hopefully he wasn't just being sarcastic perhaps he can respond to this post with some guidance or insite to clarify what he would want in that person. Also possibly sticky the post so it can gain some ground.

I figured if people got to working on it and sharing ideas and progess here for the others willing to help it would help actually getting it done instead of just wasting time complaining about it.

My intentions are an attempt to turn all that manpower into productivity.

I appreciate your interest

[Galahad]

I highly doubt the problems they have have anything to do with actual coding errors. I think it more has to do with errors in the formulas they are (or are not) using. Things like: what chance does having 75 block give you to actually block an attack, etc.

On the logic side they are probably missing info like: what is the appropriate distribution of the DIs over certain attack/defense values. The math behind this stuff isn't too difficult, and if they've come as far as they have on their own I doubt they'd have much trouble implementing it.

The big thing I think they are missing is reliable data. Until they get reliable data, what's the point of fixing anything? Replace incorrect mechanics with more incorrect mechanics?

[Dolalin]

Here's a good summary of the attack/defense system in EQ:

http://eq.knowledgepit.org/wiki/Melee_attacks

So, first we need to define the problem. At lower levels it seems mobs are hitting for more than they should, or rather, the DI is rolling high compared to what it 'ought' to be, given a player's AC/def. Max hits by mobs land too often, min hits too seldom.

Next task should be quantifying this. This involves parsing, obviously, and is much easier for someone with GM powers to do, since they can change levels / summon gear easily, so I'd recommend doing it on a test server.

Multiple classes.
Levels 1, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50.
AC: Naked, +100, +200....to softcap.
Mob: low green, blue, even, yellow, high red.
500 swings of combat or so.
We can pick a few values of Level/AC/Mob, and test the effect of agi, and see if it's in-line with expectations.

That should be set up in a nice table, organized by DI and max/min hit %. It could probably be done in a day or two.

(This is bringing back memories of my job in QA...)

Next, knowing the state of things, we should compare this to classic, or even live-EQ parses if no classic ones are available, to see what needs to be adjusted.

Adjustments can be as simple as tweaking AC/ATK values on mobs and players, or as thorough as a new attack/defence formula for the lower levels. Once the evidence is in hand, a decision has to be made here, and I guess that's where our friendly devs come in.

[Zithax]

Originally Posted by Galahad [You must be logged in to view images. Log in or Register.]

I highly doubt the problems they have have anything to do with actual coding errors. I think it more has to do with errors in the formulas they are (or are not) using. Things like: what chance does having 75 block give you to actually block an attack, etc.

On the logic side they are probably missing info like: what is the appropriate distribution of the DIs over certain attack/defense values. The math behind this stuff isn't too difficult, and if they've come as far as they have on their own I doubt they'd have much trouble implementing it.

The big thing I think they are missing is reliable data. Until they get reliable data, what's the point of fixing anything? Replace incorrect mechanics with more incorrect mechanics?

What's a step into the dark? It can't be in the wrong direction.