Working Parser?

**Motenten** · 2010-05-26 01:18

Release 1.5.6. No debuffs by time plugin yet.

1.5.6 (May 26, 2010)
+ Add Hunter's Roll to accuracy buff section of Buffs by Time plugin.
+ Add Samurai Roll to tracked buffs on Damage Graph.
+ Put pets back in damage graph.
* Change option from 'collective' to 'individual' in Damage Graph, to make purpose of option clearer.
- Fix duration of Dia II for Buffs by Time. Should be 2 minutes.
- Fix issue of mobs not being properly copied/identified in certain fallback circumstances. Should remove issue of mobs being listed in the Deaths tab.
- Fix bug in parsing status removal messages. Older parses will need to be reparsed.
- Fix bug where Benediction didn't allow for multiple targets being hit. This fixes certain instances of charmed players and/or mobs that shouldn't exist.
- Fix bug in Damage Graph that occasionally caused an index out of bounds error.
+ Add ability usage counting to the Defense tab. This allows you to see how many times the mob(s) used TP moves, even if the moves were absorbed by shadows or stunned and thus not reported in the Damage Taken section, or if the moves did no damage at all (self-cast buffs, etc). Also useful for getting a count of moves that hit multiple targets which make it difficult to get a count from the damage section. Note that the code is not optimized, and will slow down processing on this tab a bit.
- Fix parsing of Sub-zero Smash spell.
* Status removal spells are no longer listed on the Buffs page. They've been moved to the Recovery tab.
* Increase spacing on Killed By column of Deaths plugin for longer mob names.
- Fix bug of Defense tab not reporting all damage taken.
+ Add reporting of estimated MP and TP spent on curing.

**Stromgarde** · 2010-06-04 18:14

Does anyone know of a parser that will read windower logs?

**Morpheu** · 2010-06-05 20:15

Hi, sorry if this already being answered, but i looked on all pages and didnt found a problem like mine.

I just installed windows 7 on my PC, everything is working fine. The Kparse start and all, but actually dont show any data.

Can anyone help me? please

**Morpheu** · 2010-06-05 20:20

Ok, now is working, I had to actually zone for it to work. If that is normal, I never noticed before.

**Khajit** · 2010-06-10 21:06

Kparser broke. I might be because my laptop installed a java update/

Spoiler: show

**Motenten** · 2010-06-11 02:55

Looks like the config file got corrupted or something. Go to the location specified in the first line of the error and delete the user.config file. Restart KParser and update any prefs. Alternately, if it has the same problem (due to trying to upgrade from a broken previous config file), just delete the 1.5.5 (and any previous) folders.

**Motenten** · 2010-06-15 14:19

Version 1.5.7 released. Notes:

1.5.7 (June 15, 2010)
+ Addition of MP/TP efficiency calculations to Recovery tab.
* Improve processing speed of Recovery plugin. (~5x faster)
+ Add average time between weaponskills to WS Rates tab.
- Fix bug in parsing Shinsoku's additional effect: drain TP messaging.
- Fix bug in calculating MP spent on curagas.
- Weaponskill frequency by TP return should show frequency per weaponskill, rather than frequency per all weaponskills. Corrected.
* Improve processing speed of WS Rates plugin. (~8x faster)
* Refueling added to haste buffs in Buffs by Time, as well as included in Damage Graph.
- Fix issue of Fighter's Roll not being properly tabulated for Damage Graph.
- Fix bug when viewing Deaths plugin with a subset of mobs.

**Cream Soda** · 2010-06-22 00:53

menlock?

**Motenten** · 2010-06-22 13:40

1.5.8 released.

1.5.8 (June 21, 2010)
* Improve processing speed of Buffs By Time plugin.
+ Add new weaponskill, spell, and JA names.
* New memloc location: 0x005827D8

**Cream Soda** · 2010-06-22 13:43

disregard my post in the other thread asking for new memloc, lol

**Motenten** · 2010-06-22 13:46

Yeah, sorry

Got caught up in various party stuff last night.

**Darkei** · 2010-06-30 08:24

hmm I try install Kparse but I don't have

Microsoft .NET Framework 3.5 SP1
SQL Server Compact Edition

ant the microsoft.com links no work

**CDF** · 2010-06-30 17:15

Kparser Extra Attacks should be working correctly, right? In particular, I want to point out an example of where one can attempt to determine the multi-hit distribution of the Magian OA2-3T H2H.

The originator of this data set ("Bytemeee," yeah, I already noticed you gave an interpretation of the data) claims that the count data are based on H2H, yet given 573 attack rounds, the minimum number of hits without any KA, DA, etc. exceeds the actual total (1146 vs. 1142). Also, summing the multi-hit counts should still give 573 given H2H and base number of hits = 1.

I have never seen this problem (ran a "control" for H2H), so maybe people are using an old version of Kparser or just screwing up the tabulation somehow, because these results seem rather suspect, a.k.a. GIGO.

But suppose that the base number of hits was actually supposed to be 2 and the data are valid (+1 => 3-attack round, +2 => 4-attack round, up to +4 => 7-attack round). Bytemeee also mentioned 5/5 KA merits, so that should be a 17.5% kick attack rate (a "known"). Now, if 3 attacks per fist is possible (in accordance with the multi-hit distribution of the base weapon), then 7 attacks (including the kick) maximum is also possible (kicks occurring independently of punches). This is a very simple "model" of how people imagine the weapon to work.

With all that, a set of MLE estimates for the multi-hit distribution proportions are .61008282 for 1 attack, .36646327 for 2 attacks, and .02345391
for 3 attacks. But a goodness-of-fit test must be done to make sure the (extremely simple) model provides an adequate fit to the data, and clearly it doesn't, but this is more likely a situation of the data itself being garbage for whatever reason.

**Motenten** · 2010-06-30 18:01

@Darkei: I just checked the links in the readme file and they still seem to go to the appropriate Microsoft web pages. Not sure what's going on for you.

@CDF:
The Extra Attacks section should be working correctly, with one small caveat: relatively small parse samples are likely to trip up a legacy check for older parse sample intervals (1 second intervals instead of the current 0.5 second intervals). That's being removed in the next release.

Even so, testing with the EA plugin needs to be very carefully controlled. A sample from a campaign battle should be taken with a very large grain of salt.

Also, you're missing an element of that data. The "Attacks/Round" was set at 1. Therefore the +1 rounds had 2 attacks, the +2 rounds had 3 attacks, etc. You can't just reinterpret those as a normal 2-attack round and then add the +1/+2/etc extras.

Originally Posted by CDF

yet given 573 attack rounds, the minimum number of hits without any KA, DA, etc. exceeds the actual total (1146 vs. 1142). Also, summing the multi-hit counts should still give 573 given H2H and base number of hits = 1.

The data provided:

1142 melee hits
1 attack/round (so +1 is 2 hits, +2 is 3 hits, etc)
573 rounds, distributed as:

373 +1's = 746 hits
50 +2's = 150 hits
19 +3's = 76 hits
6 +4's = 30 hits
3 +5's (assumed) = 18 hits

746 + 150 + 76 + 30 + 18 = 1020
That leaves 122 1-hit rounds to get 1142 total hits.

373+50+19+6+3 = 451, matches the number of multi-attack rounds reported.

573 - 451 = 122, number of 1-hit rounds; matches hit count.

I'm not sure how you got 1146 for any total.

**CDF** · 2010-06-30 18:22

For my "control," I had 84 melee rounds and 192 attacks. These totals should be invariant regardless of the base # of attack rounds.

Setting my base # attacks to 1 (wrong) gives 108 extra attacks, with 60 +1 rounds and 24 +2 rounds (no +3 rounds, as it was a small sample). Knowing that the base was wrong, I can interpret the number of "+ 1 rounds" as 2-attack rounds (two swings always occurring, obviously) and the number of "+ 2 rounds" as 3-attack rounds (where a DA occurred).

That is absolutely consistent with the output when the base # of attacks = 2.

In any case, for a H2H situation the number of melee attacks should always be >= 2*(# melee rounds, equal when no DA or kicks are possible), and the data set I referenced fails that check because 1142 < 2*573 = 1146. And the guy himself said he was checking hand-to-hand later in the thread (not Footwork).

In any case, that data set is completely unacceptable on statistical grounds so no interpretations should be made from it. I did my analysis based on 122 2-attack rounds, 373 3-attack rounds, etc...

Of course, one can ask why the attacks/round was = 1 to begin with...

**Motenten** · 2010-06-30 18:47

Originally Posted by CDF

In any case, for a H2H situation the number of melee attacks should always be >= 2*(# melee rounds, equal when no DA or kicks are possible)

Not strictly true. You also have to account for melee rounds that killed the mob with less attacks than that (ie: killing with the first punch of the last round). However that only reduces the number by 1 in this case.

Also, while you are correct in how you can reinterpret data that was originally 2/rnd as 1/rnd and figure out what the original numbers must be, shifting this from a 1/rnd to 2/rnd leaves you with the untenable conclusion of there being 122 rounds where you never punched a second time. While you may occasionally get that from killing the mob, there's only 1 instance of that in this data set.

One must therefore conclude that he was using Footwork for at least a fairly large portion of this parse, in order to get 1 attack per round so frequently. It also means that it is nonsensical to expect the total number of hits to be at least twice the number of rounds.

**CDF** · 2010-06-30 20:01

I concede that I didn't consider "censored" attack rounds (this is at worst a minor confounder though given the scenario), but to make an empirical determination of the multi-hit distribution, which is the putative goal of getting the data in the first place, precludes doing crap like running kparser in the middle of Campaign or, worse, a merit party. Also note that my naive criterion is a lower bound. Obviously with a multi-hit weapon, the number of attacks should be significantly higher as it's not like he one-shot mobs left and right.

If it really follows that he was using Footwork, then analyzing the data is utterly pointless, and one does so only to underscore the "fact" that the count data is B.S.

Edit: I followed up on that abortion of a thread and I noticed some dipshit referring to a blog without actually citing it (providing a link: here) as though he wanted to remain privy to some exclusive knowledge (dumbass fuck). Anyway, some of the reported estimates of the OAT rate, which people concluded was 40% (who knows if they even know how to count, though) suggested it could be higher than 40%. Of course you got one THF (TA anyone) rocking a Brutal and some other stupid crap. Rather suspicious.

**Hanyoko** · 2010-07-01 09:58

Originally Posted by Darkei

hmm I try install Kparse but I don't have

Microsoft .NET Framework 3.5 SP1
SQL Server Compact Edition

ant the microsoft.com links no work

windows update should get you .net 3.5 but if not you can try their Product page and get the assisted downloader.
And from that page you can get the Full package installer which is approx 300mb

The download assistant only gets what you need and requires an active internet connection to install, whereas the full package can be considered an "offline install" as it contains everything.

**Synched** · 2010-07-05 20:32

Problem fixed, thanks.

**Motenten** · 2010-07-05 21:45

According to Microsoft at http://www.microsoft.com/typography/...aspx?FMID=1805 the Tahoma font should be part of the OS installation. I'd try reinstalling the font from your OS install disk.

Thread: Working Parser?

Thread Tools

Similar Threads

Working Parser

Working parser?

Can I get a link to a working a good Parser?

Latest Threads

Up & Coming Threads

Hottest Threads