Fight against unkillable level 42 named in Vi'rak lair (roughly a million hp, self heals 16% a pop, and then runs at 10%, 1% per 2 seconds or so regeneration with that hp total is just wrong) - provided some decent parsing material though. This is with dog shaman, disciple, ranger and psionist buffs/debuffs. One Secret of Celerity was used.

Total Hits:      1004
Total Damage:   778201
Lowest Hit:      0
Highest Hit:      6517
Total Time in Combat:   1823
Average Swing Time:   1,815737
Damage Per Second:   426,8793
Damage Per Hit:   775,1006

This looks like the issue where the client logs some un-godly huge damage number.  I'll add a check to reject these entries - as they are clearly wrong.


The short answer is yes - that would be counted.  Since I was interested in comparing different players and NPC's, and you do not see their attack on/off messages, it was clear that a different approach was needed.

Each consecutive attack is logged with a timestamp of when it occurred.  The length of seconds that any given action "takes" is the delta (in seconds) between it and the previous attack.  The problem with this is the last attack in any sequence takes much longer than any other.  Say you finish a mob - the time between that and the next attack could be seconds... minutes....hours...who knows.  So, I use a "maximum attack delay" of 3 seconds to fudge any entry with a duration longer than that amount of time.

While this is not a perfect solution, it does allow me to track a number of things I could not account for given the content of the Vanguard log files - such as NPC DPS, other player DPS, caster DPS (since they do most of their damage out of combat,) etc.

No 517428174281478 hits in that fight - notice the fight time (30 minutes?), total hits and highest hit.

[/quote]

Sorry, are those commas or periods?  It may be a localized windows thing - but it looked like the DPS was more than 400k.

Suggestion:  Can the parser be set up to count the number of dodges and parries, and blocks for sake of completeness for other classes?  These should probably be reported as a number as well as a number / unit of time.

Or perhaps instead of a number / unit of time, they could be reported as percentages.  But the percentages should be based on the order in which the avoidance moves are calculated.  For instance if the first calculation is hit/miss, then block if it would have hit, then parry if it would have hit and not been blocked, then dodge if it would have hit and not been parried or blocked, the formulas would be

misses% = misses /  totalAttacks
block% = block / (totalAttacks - misses)
parry% = parry / (totalAttacks - misses - block)
dodge% = dodge / (totalAttacks - misses - block - parry)
totalAvoidance% = (block + parry + dodge ) / (totalAttacks - misses)

Note that hit% is NOT 1 - misses% because attacks that would have hit still have a chance to have been blocked, parried, or dodged.

To do this, we'd have to find out for sure what the order is for checking to see if an attack that would hit is avoided.