Working Parser?

[Bakthi]

I have the same issue as Thunder, and ran the same tests with exactly the same results (parsing from live data still didn't work after parsing from the log, which did work). 1.3.6 still works perfectly well; 1.4.2 didn't, and 1.4.4 doesn't. They're all using the same offset, all running as Administrator under 32-bit Vista. FFXI is running in a window with a border; I don't full-screen it.

I am using the option to select the FFXI process, but I also tried it without this option and nothing changed (I'm only running one FFXI instance at the moment).

I'm also still having the Ouryu issue. I parsed our fight last night (with 1.3.6). Under "Fights", it does list Ouryu as having been "defeated", but no player name. When viewing the data in 1.3.6 in the Offense tab, trying to select Ouryu throws up a Windows error. When doing this is 1.4.4, it simply does nothing; the Offense tab continues to show whatever it was showing before. I can isolate damage for any of the Ziryu or elementals, but not Ouryu.

PS: I love KParser, and am very grateful for it. Just frustrated by these bits.

[Motenten]

Bakthi, could you send me a copy of that Ouryu parse? Save a copy without chat info if you wish, then either attach it as an issue to the google code site, or send to the [email protected] email.

Still trying to figure out the parsing problem.

[Bakthi]

I submitted an issue with the file attached; thank you! Oddly, I have two parses with this behaviour and one that works fine. I'll be happy to provide the others later if necessary.

[Motenten]

New memloc is 0x005801d8

Bakthi, I'm trying to work a solution for that that doesn't involve hard-coding the names of most the NMs in the game. Still having trouble with it, but hoping to have something out tonight.

[Motenten]

Ok, new version out: 1.4.5. I believe it fixes Bakthi's problem, as well as a couple other issues that have come up. Has the new default memloc in place. Regular zip file as well as the installer version are both available.

Ranged attacks are not completely updated with today's patch, but they at least get the normal and sweet spot attacks. Will update with any other messages once I find out what they are.

Changelog:

1.4.5 (November 9, 2009)
* Made changes in analyzing fight deaths. If an 'other' entity (non-self/party/alliance) kills a mob, the program firsts checks to see if that entity had any other actions against the battle in question. If not, then it enters it as a separate fight instead of ending the current fight.
The purpose of this is to fix the issue of someone outside your party killing a mob and it ending your current fight, showing up in the mob listings as (for example) two mobs killed, each with half the normal health, one for 0 xp and the other for normal xp. This causes problems in segregating test data by mob level.
The requirement for the 'other' entity to have performed a visible action in the current fight is to not erronously create fake fights due to pets (avatars/puppets/etc) getting the killing blow, since all pets are classified as 'other' entities. If your filters prevent you seeing such actions, you may end up with misleading fight data.
+ Offense and Defense Details plugins now show an integrated sequence of melee and ranged hits (crits and non-crits) when Show Details is selected.
- Fix bug in displaying Raw Data when switching between all and unparsed.
- Improve handling of identification of Notorious Monsters as mobs rather than charmed players, and the consequant occasional mis-identification of players as charmed mobs.
+ Add parsing of new "sweet spot" message for ranged attacks. Still missing the third possible ranged attack message.
* Updated default memloc to new value: 0x005801d8


Edit: 1.4.5.1 released with the full fix for the new ranged attacks messages.

[Kohan]

Despite having no filters on, whenever I use KParser, I appear to be missing quite literally half of the total damage done.

Do you have any idea why this happens? It's normally only when something very fast (e.g., a zerg rush) is done, but it's occurred during longer events like Dynamis, too.

As a side note, although I am currently operating with only 2GB of RAM, I used to have 4GB, and had this problem then as well (if this has anything to do with it).

[Motenten]

I'm not directly sure, but I can guess on a couple things.

Under ideal circumstances, the parser should be able to keep up with ~100 lines of text per second. The chat log buffers update once per second (though the parser checks them every half second), keep 50 lines of text in a current buffer, and keep the 50 lines before that in another buffer. Anything older than that is dumped from the direct chat area. If more gets sent through than that, there's no real way to avoid data loss.

The other thing that can slow things down is having too many plugins open at the same time. Each plugin takes a certain amount of time to process, and that can impact other processes going on (depending on computer speed). The parsing and the updating are done on separate threads, so it -should- avoid blocking the ability of the monitor thread to read more chat data, but it's kind of hard to completely guarantee that. To get maximum parsing speed, close -all- plugin tabs while parsing.

[Masamune]

Do you mean Kparser plugins, or windower plugins ?

[Bakthi]

Awesome! I had to have it reparse the data, but Ouryu separated out perfectly. Thanks a ton.

Unfortunately, so far it still won't parse from live data for me, but I've only tried a little. So much update content to work on.

[Motenten]

KParser plugins. Under the Windows menu, select Close All Tabs.

[Cream Soda]

Memlock for Dparse? Kparse is working again but mem posted here not working for Dparse.

[Motenten]

DirectParse memloc should be exactly the same as KParser's. Not sure why it wouldn't be working for you...

[Cream Soda]

Tried entering it in twice, no go.

[Shosei]

I'm having the same issue. I tried using the new memloc with DirectParse and couldn't get it to work.

[avelle]

MEMLOC=0x005801D8 works just fine for me.

[Cream Soda]

MEMLOC=0x005801D8 works just fine for me.

Case sensitive now? I had the same thing but lowre case and it didn't work. Tried urs and it works.

[Ksandra]

ok this is probably a very stupid question, but I can't see my friend's screen so here goes.

Friend asked me to find him a parser, I had him d/l kparser, took a bit but I guess he has to put in that new memloc code? He said he tried but every time he does it goes back to the default code or something. Anyone have any idea what he might be doing wrong?

[Motenten]

If he downloaded 1.4.5 (and isn't upgrading over an older version) he shouldn't need to change the memloc at all. If he were upgrading over an older version, it would carry over the old memloc and he would need to change it. If he downloaded 1.3.8, he will need to change the memloc as well.

As for actually changing the value: click the Edit button, input the new number, then OK out of the dialog box. Settings are saved in a 'safe' location; it doesn't matter whether he has administrative privileges or not (though you'll need to have those privileges to parse from RAM, so probably will have them anyway), so there should never be any problem with being denied write access and ending up with the new number not sticking.

If that simple set of instructions fails to work, you'll need to get more information from him: the exact set of steps taken, any errors in the log file, and any warning messages from the OS.

[Motenten]

Version 1.4.6 released.

1.4.6 (December 2, 2009)
- Fix issue with filtering out 0 XP mobs when 'All' mobs were selected.
- Fix issue in selecting certain subsets of mob groups.
* Rewrote backend code for Extra Attacks plugin. Now *much* faster (30k attack test parse for Zanshin took 75 seconds before; now takes 370 ms), and I believe more accurate in determining the separation between multiattack rounds.
- Fix display bug in Performance tab that was the result of improper rounding of time values.
- Check for item names that exceed the database size limit.
- Fix potential deadlock in database code.
+ Provided more status feedback if Debug Mode is on.
- Fix problem of losing track of chat log in RAM if player changes characters.

[Ropeydonkey]

Version 1.4.6 released.

1.4.6 (December 2, 2009)
- Fix issue with filtering out 0 XP mobs when 'All' mobs were selected.
- Fix issue in selecting certain subsets of mob groups.
* Rewrote backend code for Extra Attacks plugin. Now *much* faster (30k attack test parse for Zanshin took 75 seconds before; now takes 370 ms), and I believe more accurate in determining the separation between multiattack rounds.
- Fix display bug in Performance tab that was the result of improper rounding of time values.
- Check for item names that exceed the database size limit.
- Fix potential deadlock in database code.
+ Provided more status feedback if Debug Mode is on.
- Fix problem of losing track of chat log in RAM if player changes characters.

Sweet - good work^^

I will download it the moment I get home and test in Limbus later