Pylon Simulator

[Dismiss]

Before anyone gets too excited, I'd say this is at an alpha level stage. I have a lot going on in my personal life right now, so I doubt this will be ready any time soon.

 

I'm trying to make a pylon simulator to make testing ideas easier. It's a lot easier to draw a solid conclusion with 3000 actual runs than a couple times that are subject to a lot of variance. Basically the program right now lets you set up the skills you're using (only care about anything with offensive ability here as I don't model incoming damage), define how you prioritize your attach chain, and then runs a sim to graph a PDF and CDF of the Monte Carlo run.

 

Where I've gotten so far is thanks to @modest, @Bopper and @RubyRed.

 

I think I have most of the basic sets working okay for the melee classes, but don't have data to test yet. By basic I mean ones without power redirect (like Titan Weapons with different animation times) or psuedo pets. I have handled some specific cases of these, but want to rework how I'm doing that to hopefully make it more generic and save me time in the future. I think it's possible basic sets for ranged classes would work fine too.

 

There is something I would love help with to know I'm headed the right direction: data.

Specifically, I need actual pylon times for some more sets. I don't have a lot of time to do the coding, so it would be very helpful to farm that out.

 

So far, with Modest's help, we have lined out a Rad/claws tanker (in the gif), Fire/shield scrapper and savage melee/energy aura Stalker.

 

What I'm looking for is this:

• data from melee ATs (and possibly blaster/defender too)
• the mids file for the build
• 10+ pylon times are really needed for me to judge if my results look good or not
• the only Destiny I have modeled right now is Ageless T4, although I could model lower tiers. If you are using clarion +special for shenanigans, that will have to come later.
• nothing other than assault hybrid. I strongly prefer if it's left on passive for the runs, because I haven't modeled active Assault yet, but if you already have a set of times with it active, it could be done.

 

If you have the above, you can post it here or I'm happy to talk you on discord at Dismiss#5787.

 

If there is someone interested in helping code this, I'm open to it since it's a lot for one person, but you'd need to be familiar with JavaScript. Warning: I am not a programmer by trade, so if you are and want to help.... I'm sorry, I'm sure there is stuff that is not proper.

 

@UberGuy I've also heard you've updated Ruby's API for the latest versions for your cityofdata site, so I'd love to talk to you about that too.

 

[UberGuy]

While I've kept parts of Ruby's tool up to date, I don't use it like most people would - my build is invoked via an internal API wrapper that I call from within Python code, and emits data directly to Python. I haven't been updating the export code that is how most people would use the command-line tool. Basically my build is a fork that's diverged from Ruby's version, and Ruby's version now has more "raw" data export features from when I first forked off of it, which was a big reason I forked it to start with.

 

That's a long winded way of saying my updates really aren't such that you could use them directly.

 

I have two intended ways of addressing this.

• I've long intended to contribute updates back to Ruby's mainline code. This shouldn't be too hard, but I've let our codebases diverge enough that it's going to be a bit time consuming for me.
• I also intend to provide a download from CoD of an archive of the "raw" version of the game data that my Python code produces. This is (very) similar to Ruby's version, but lightly optimized for how I build the "final" version that back's CoD's displays.

My focus for a long time was just getting CoD itself working, displaying power, AT and critter data fully and correctly. Then I left it alone to focus on my job and then take a vacation. I'll be back soon (days) and should be able to work on both.

Edited August 9, 2021 by UberGuy

[SomeGuy]

I've got a lot of times with different AT combos/builds I can send your way. I just know I don't have a new time for post-TW nerf (it's still around 1:30 something) or post-snipe nerf. Those builds had about 10-20s added to their times. Nothing huge. EM/BIO is DEFINITELY the fastest melee pylon killer I have right now though.

[RubyRed]

On 8/8/2021 at 2:05 PM, UberGuy said:

While I've kept parts of Ruby's tool up to date, I don't use it like most people would - my build is invoked via an internal API wrapper that I call from within Python code, and emits data directly to Python. I haven't been updating the export code that is how most people would use the command-line tool. Basically my build is a fork that's diverged from Ruby's version, and Ruby's version now has more "raw" data export features from when I first forked off of it, which was a big reason I forked it to start with.

 

 

I apologize that I've not had time to keep my code up to date with the latest changes.. the Homecoming team makes structural changes to powers.bin far more often than I imagine the retail team ever did, and the reverse engineering takes a lot of time and effort. So I kinda signed myself up for a sisyphean task. 😛  I've had a lot of other IRL stuff come up this year that's kept me away from the project.