Unless they changed something evasion checks first, so you don't even have a chance to parry if you evade. It shouldn't be in the denominator.Originally Posted by Hakamaru
Unless they changed something evasion checks first, so you don't even have a chance to parry if you evade. It shouldn't be in the denominator.Originally Posted by Hakamaru
Right... but it can go either way depending on what you're interested in. If you want to see how much your parrying is doing for you in a raw sense, you'd want (numParries/totalSwings). Now, if you're only interested in true parry rate considering the evasion precedence, then you want to see (numParries/totalSwingsNotEvaded).Originally Posted by DrObvious
I think I'm going to agree with you here and choose the latter. I've fixed a few small issues surrounding this, like the parry column wasn't getting updated when I lost an Utsusemi shadow. With my evasion build on EPs, the parry column would sit there untouched for really longs periods of time.
hits for 0 should be included as a hit (for accuracy purposes) and should probably be included in the average damage per swing as well.
basically, as a hit -
as for parry; what about adding a column to report both, similar to counter rate?
Fixed the hits for 0 things late last night.Originally Posted by Amele
There's another dev I haven't seen post here that's tearing through stuff and actually beat me to the basic evasion fix (yay conflicts). They've also fixed loot parsing, Magic Bursts, and added a WS count as well as some other minor additions (third eye didn't used to register as an absorbed hit). Just need to do some more testing and then we should release a binary version.
Originally Posted by Hakamaru
she fixed all that other stuff already? damn lol
hurrah for multiple school projects
For parrying, it may be good to include two options, one giving you your raw parries/mobatks, and the other removing evades (but not shadows) from it, so that you can see both how it's doing for you as a whole and what your base parry rate is, kind of like how the Counter parser has two options for raw and base.
we probably shouldn't be including the hits for 0 in average swing damage. Two sides to this:
gimps who truly hit for 0 because they suck and
trolls/cerberus stoneskin
the latter will skew people's results on trolls by a lot, and the lowest hit, unless we exclude 0s, will be 0. Average dmg/swing will also suck.
I disagree. Capturing the 0 hits is a more accurate reflection of reality. That and it helps debugging evasion issues on lvl 0 rarabs that hit me for 0. >.>;;Originally Posted by Zhais
If something hits a RDM for 0 when he has stoneskin up, it should be reflected in his evade rate, since he didn't evade that attack. It should also go into a melee's ACC rate as well, since a hit for 0 is still a hit.
zhais isn't saying that it shouldn't count as a hit; what he is saying is that is shouldn't be applied to the damage averages.
the issue is, I think that both values use the same underlying 'hit counter'
we could probably add another counter that tracks hits for 0 and subtract from base hit total for calculating average damage, etc.
What if you are actually hitting for zero?
That's one of the things I mentionedOriginally Posted by Ferien
One thing that might solve it: removing any statistical outliers from the damage summations. That would hopefully remove a lot of the 0s from stoneskins... and if someone is bad enough that they hit for 0s without stoneskin, those 0s just may be within the range to not be considered an outlier.
We wouldn't remove those for accuracy/evade etc. rates, but for damage averages, minimum and maximum.
Can you just capture Diamondhide/Stoneskin/Hoplon events and ignore all zeros until the next >0 hit?
Ok, I just uploaded a new release of the code. Everyone that requested changes really owes a big thank you to Releena for getting most of these changes / fixes taken care of in such a timely manner.
http://sourceforge.net/projects/directparse/Originally Posted by Change Log
As for the hits for 0 thing... It would be very simple to ignore 0 hits in regards to minimum damage tallies. However, it's a larger change to get them to not affect average melee numbers. As Amele said, you basically have to record them as a new class of melee damage, then account for this new class back in all the defensive & accuracy stats. Not terribly difficult, but didn't make it into this release.
Generally you'll start hitting for 0 before the effect shows up in the log. Depending on how the parser works (or how the game sends info to you, even), it could be impossible to pick up the event exactly when it occurs.Originally Posted by Delekii
I might have missed it somewhere in the setting. But does/can the parser keep track of the mob level? I remember seeing somewhere in here that mob level was something that is sent.
Edit: As well, I know this could be a freak coincidence... but, does the code for the loot effect anything in game, or for that matter does any of the code in anyway effect the game?
The reason I ask is, lets say, my farming rate was 1/8 (Average over months of testing). As soon as I downloaded and started using the parser, I went 0/24 before I had to log.
Edit2: As for the mob level, if it is possible. Could (or how do I set) the parser somehow split the parses by mob level?
I vote for bad luck on your part. 0/24 is well within the standard deviation of a 1/8 drop.Originally Posted by Repairs
However, if the parser can cause 0% drops, please invert that code so we can get 100% drops.![]()
It does not, and the parser would have to change quite a bit in order to display that in a meaningful way. You'd need to break apart parses into per-mob sub-parses, and there's nothing like that in the code.Originally Posted by Repairs
Nope. It simply reads the log out of FFXI program memory space. Nothing in the game memory gets altered as far as I can tell.Originally Posted by Repairs
exact mob level is sent in some packets (notably the /check packet ; dunno if it's sent in others) but it would take a pretty drastic change to the application code to both get and display that in a meaningful way.
there's been talk on and off of a more detailed heads up about mobs, incorporating a database of mob information and some info skimmed off the memory map - but that's fundamentally different than a parsing application.
Ya, 0/24 was just really bad luck... everything went back to normal.
I hope you all can figure out a way to get the mob level incorporated. Seeing as the stat's the parser provides are great, I would really enjoy even more accurate numbers! Things like trying to test palwan vest vs assassins vest, etc.
Over all, a great job by everyone! thank you
how would telling you what mob everything you fought was, in a 300 page document, make the parser more accurate?Originally Posted by Repairs
if you're trying to control your results, you can do that independently (by assessing mob level yourself) and a parser is really only useful for getting an independent average for melee, and nothing else, in that situation.
if you're wondering which is better for meriting or hnm or w/e... well. I would think you would *want* level variance in your data. *shrugs*