 |
10/01/12, 2:10 AM
|
#511
|
|
Soda Popinski
|
Yeah, I'd just add those stats in with the manual adjustment for now. Good luck with drops!
|
|
|
|
10/01/12, 6:42 AM
|
#512
|
|
Great Tiger
Night Elf Druid
Echo Isles
|
NG Uptime refinement?
I know this is a pain to do in a single pass, but as-downloaded on the Rotation page I see:
Final Time: Lunar=11.2, Post-Lunar=5.4
I'd expect post-lunar NG uptime to be close to 1-(11.2+5.4-15)/5.4=70%, but the spreadsheet reports (rotation!e22) 50%. Perhaps that number contributes to high Haste valuations?
Something else to consider: An in-progress cast doesn't care if NG falls off, and I suspect simulationcraft effectively finds additional haste breakpoints where more nukes get NG. For the Rotation sheet you might want to say the NG duration is something like 15+SFNGCast/2 (or WNGCast/2 for solar/post-solar). Edit: Just this change drops haste rating from 2.02 to 1.94 on the as-downloaded sheet changing
NG Uptime Expanded (Eclipse) to (15+SFCast/2)*1.15, and NG Uptime Expanded (Solar) to (15+SSCast/2)*1.15
Last edited by Erdluf : 10/01/12 at 6:49 AM.
|
|
|
|
|
10/01/12, 10:29 AM
|
#513
|
|
Soda Popinski
|
Quick update with a few bug fixes:
- Armory sometimes giving level 60 and 90 talents round the wrong way. Now checking for talent names rather than trusting the number given.
- Spellthread IDs swapped around for import.
- Belt buckles should now pick up properly.
- Umbrella of Chi-Ji's stats updated to live values.
- Bracer Intellect enchant's value corrected.
To be done in the next version:
- Fix profile saving/loading.
- Add T14H BiS (sorry, I keep forgetting to do this).
Last edited by Tecton : 10/01/12 at 10:49 AM.
|
|
|
|
10/01/12, 10:43 AM
|
#514
|
|
<Druid Trainer>
|
On the second point: yeah, I can see adding those half-casts to the NG uptime. I should think about why that reduces value of haste and make sure there's no DoT clipping going on.*
Modeling it more detail is probably not worth it since (and this has come up before with timed Eclipse in WLK), I'm not really sure how rigid those breakpoints are in practice--in any event, don't think people are going to gear around them. And if people don't gear around them, there isn't too much point having a fancy system that breaks out all the haste stuff like we do for DoTs. The long-term average scaling will still be correct.
*Related: It might almost be better to assume, slightly unrealistically, that you get one full Eclipsed DoT and one full uneclipsed DoT per cycle. If cycle length at BIS haste right now is between 31 and 32 seconds, and the two DoTs together take just over 32 seconds, this would only overvalue DoTs very slightly and is probably better than trying to clip off that one tick, which interferes with scaling. [e: forgot about crit extensions. Will have to think about this.]
------
The first point is trickier. Your estimate of what NG uptime should come out to still isn't quite right, because I'm using "uptime" a bit misleadingly--it's really a percentage of casts affected by NG, not a percentage of time. Note that the only two places NG uptime is used is the final time spent casting Nukes and Starsurge (rows 25 and 27). So that's why NG uptime is based on the pre-NG time. But, that doesn't account for the whole discrepancy. There's an added deviation from the fact that after NG uptime is computed, the length of the phase is further reduced by instant SS procs, and that deviation is real.
I'm inclined to leave as is, since 1) trying to anything more would be a giant mess like in Cata, and 2) this is all sort of approximate anyway since it doesn't account for the NG time you use casting DoTs and things (which would reduce the percentage of nukes affected by NG).
Last edited by Hamlet : 10/01/12 at 11:41 AM.
|
|
|
|
10/01/12, 12:09 PM
|
#515
|
|
<Druid Trainer>
|
Here's an update with
--Windsong proc rate reduced by factor of 3. Current model has it proccing once per minute, increased by haste.
--Div/0 and negative NG uptime issues fixed.
--Added Erdluf's suggestion of slightly increasing NG uptime.
I've not changed the modeling of the two-DoT cycle for now. With crit extensions, fitting a full extended Eclipsed DoT and a full uneclipsed DoT (total duration about 34.5s) inflates DoT DPS a good bit. The problem with leaving it how it is (full extended Eclipsed DoT + fill in whatever time is left with a clipped uneclipsed DoT) is that it devalues both haste breakpoints and crit extensions, because they don't have room to create any additional total DoT ticks. If we really did clip an uneclipsed DoT at the beginning of each Eclipse, that would be correct, but I'm not sure people are going to do that (but maybe that's what would happen over a long stretch with no movement).
So I want to figure out some way to capture the benefits of added DoT duration without assuming you can have >100% uptime on any DoT.
Last edited by Hamlet : 10/01/12 at 12:15 PM.
|
|
|
|
10/01/12, 2:38 PM
|
#516
|
|
Soda Popinski
|
Someone just asked me in game, did the 3% Starfire & Starsurge, 9% Wrath hotfix ever make it into the sheet? I honestly can't remember.
|
|
|
|
10/01/12, 2:42 PM
|
#517
|
|
<Druid Trainer>
|
I never changed anything, but what's in there still matches what's in the spell databases online. So either 1) those haven't been updated yet, 2) changes never went live, or 3) it was actually already updated when I put the values into the sheet.
Right now I have 1.027 for Wrath. If anyone could confirm that either 1) it's actually 9% higher than that on live right now, or 2) it was actually 9% lower than that a few weeks ago, that would be handy.
|
|
|
|
10/03/12, 7:37 PM
|
#518
|
|
Glass Joe
|
I got WrathCalcs working on 64bit Windows with Excel 2010. I replaced the URLEncode routine with a system agnostic routine that doesn't require Kernel32. I also replaced the JSON evaluation mechanism to use a custom parser class rather than using ScriptControl. Everything is working for me, but I would like someone to try it on 32bit to confirm.
Uploaded to Google docs. Just click the link to save the original file.
https://docs.google.com/open?id=0B34...jdEajd5VlByTVk
|
|
|
|
|
10/03/12, 9:12 PM
|
#519
|
|
Soda Popinski
|
It's returning all blank data on 32 bit, sorry!
|
|
|
|
10/03/12, 9:47 PM
|
#520
|
|
Glass Joe
|
Originally Posted by Tecton
It's returning all blank data on 32 bit, sorry!
|
Do you have special characters in your character name or server name? I'm hoping it is just the URLEncode that's failing.
|
|
|
|
|
10/03/12, 11:15 PM
|
#521
|
|
Piston Honda
|
Originally Posted by Proudhorn
I got WrathCalcs working on 64bit Windows with Excel 2010. I replaced the URLEncode routine with a system agnostic routine that doesn't require Kernel32. I also replaced the JSON evaluation mechanism to use a custom parser class rather than using ScriptControl. Everything is working for me, but I would like someone to try it on 32bit to confirm.
Uploaded to Google docs. Just click the link to save the original file.
https://docs.google.com/open?id=0B34...jdEajd5VlByTVk
|
On 64-bit, when I put in my char name/server name and set the region to us.battle.net, after I hit the Battle.net button, the spreadsheet freezes for a moment like it's updating, but then all the fields are blank. Though, it does find my ring.
|
|
|
|
|
10/04/12, 5:57 AM
|
#522
|
|
Soda Popinski
|
Originally Posted by Proudhorn
Do you have special characters in your character name or server name? I'm hoping it is just the URLEncode that's failing.
|
No, afraid not, just the details on the left.
I've tried various options in the past to get it working for 64 bit Excel too, and came up against the problem time and again of it working for one, but not the other without requiring addon installation/etc which isn't really a great solution. Considering the vast majority use 32 bit, the fact you get 32 bit for free if you have 64 bit, how poorly Excel 64 is supported and how little benefit you get from using it (I've still not seen a spreadsheet that needs the size increases, even in a corporate setting) I'll probably continue with what I have. The other thing is that I'll be honest and say I'd have issues diagnosing issues with/making changes to your code.
Couple of small changes:
- Corrected where Lightweave's stats were counted so it didn't throw off character sheet values.
- Tweaked T14H profile with the discussion in the main balance thread in mind, with slight adjustment.
|
|
|
|
10/04/12, 8:36 AM
|
#523
|
|
Glass Joe
Troll Druid
Lightning's Blade
|
On the 64/32 bit issue, I can't use the 32bit version.. "enabling content" breaks the spreadsheet.. I'm on 64bit.
|
|
|
|
|
10/04/12, 9:42 AM
|
#524
|
|
Soda Popinski
|
You're going to need to give me a little more to work on than "breaks the spreadsheet". Do you get error messages, are there calculation errors on the sheet, etc?
|
|
|
|
10/04/12, 12:34 PM
|
#525
|
|
Glass Joe
Troll Druid
Lightning's Blade
|

After enabling content, this error pops up and it opens Microsoft Visual Basic. Refreshing (f9) brings this up, using any of the battle.net features brings this up. The load/save/import/export brings this up. This isn't any recent issue, I've had it for a long time.. when it was first added way back in the day it worked for me though.
If I don't "enable content", all calculations work fine as far as I can tell. I just have to manually input gear.
|
|
|
|
|
|