The latest version seems to have a bug when calculating the ranged attack power / crit rating summary for an item.
For example, my Towering Mantle of the Hunt shows 123,1 RAP / 1,86% crit chance, though the actual RAP increase is only [48 + 30(Enchant) + 10,8(45% from Int with Careful Aim)] * 1,04 (2 points in Master Marksman) = 92,325, and the crit chance which should be 1,4% (21 + 10 Crit Rating) shows up as 1,86%.
The latest version seems to have a bug when calculating the ranged attack power / crit rating summary for an item.
For example, my Towering Mantle of the Hunt shows 123,1 RAP / 1,86% crit chance, though the actual RAP increase is only [48 + 30(Enchant) + 10,8(45% from Int with Careful Aim)] * 1,04 (2 points in Master Marksman) = 92,325, and the crit chance which should be 1,4% (21 + 10 Crit Rating) shows up as 1,86%.
I think it's counting some of the bonuses from enchants twice. I noticed the same thing yesterday when comparing my Tier 4 legs with Nethercobra Leg Armor to Tier 5 legs with no enchant.
local text = text; -- XXX Butt ugly fix by Mikk to prevent text from being a table when getting here (code below modifies it for subsequent loops... oof)
replace with:
if "table" == type(text) then
text = ""
end
StatLogicLib is... The only variable name used all over the file is 'text', so i couldn't really get the part where suddenly 'text' becomes a table in those number of for-loops :<
I haven't d/l mod yet. Does this work for every class? I have a few others that are class specific....meaning more mods..
Yes, it does work for every class. RatingBuster isn't a "points" mod. It simply converts combat ratings into percentages, and shows you the total effective stats for your class, factoring in talents. Although this is the kind of question that when I see, I think why don't you read the description, or just download it and see?
It seems to be an ambiguity problem.
In StatLogic r33977 there is a variable named text getting filled by a strsplit.
Lua obliges by making that a table and filling it with substrings.
Then later a second variable meant to hold a simple string is declared as text again in the same scope.
To prevent the errors the simplest method is to go to lines 6259 and 6260 and change the text variable to a different name.
From
text = {strsplit("@", text)}
for _, text in ipairs(text) do
Change to
textt = {strsplit("@", text)}
for _, text in ipairs(textt) do
(note the extra 't' for the table variable)
Exact same change to lines 6317 and 6318.
That makes r33977 error-free.
This that I posted in previous page is imo a better "fix" than all the tricks trying to force text to be a string.
There's a reason it's a table and the code is looping through it.
Problem was ambiguity in variable name.
Doing those changes to r33977 fixes it.
Don't know what is the current revision code, since I stopped updating after I fixed statlogic r33977.
(until Whitetooth has a chance to go over his code himself)
I haven't d/l mod yet. Does this work for every class? I have a few others that are class specific....meaning more mods..
Yes, it does work for every class. RatingBuster isn't a "points" mod. It simply converts combat ratings into percentages, and shows you the total effective stats for your class, factoring in talents. Although this is the kind of question that when I see, I think why don't you read the description, or just download it and see?
:) .....I have several other mods that do something like it for different class's
Oh, the whole reason I came to this thread, and completely forgot when I made my last post:
Could we get an option to not show the sigma graphic next to the "stat summary" header in the tooltip? I think the blank line is enough of a separator, and the icon just looks... out of place.
hi there,
great addon, love it!
minor issue but no bid deal. when i click a link to an item the stat numbers appear halfway outside the frame... if i just close the window and click the same link again it shows up just fine. i don't think this works when hovering inventory items, only when i click a link to an item, ie from the chat window.
if anyone is having the same problem, any advice is most apreciated
thanks a lot
have you tried if you get the wrong numbers without other mods messing with the tooltip ?
I could imagine that the coloring mod is adding/changing lines in the tooltip which might make ratingbuster count those lines twice.
yeah same problem here with counting stats on my Spellstrike Hood, see below
its not 112 Spelldamage its only 85
this makes it a little hard to compare items if u just rely on the compared stats and don't recheck them
that's the a bug in statlogiclib
whenever there is a line with 'and' then both or one of the stats are counted twice.
in your screenshot:
+22 dmg AND 14 hit
4 crit AND 5 dmg
The double-enchant thing is indeed a problem with StatLogicLib.
I've got a pastey of the diff of my fix, which I believe is far less hacky than others - it reduces quite a lot of the code and seems to produce correct values from my brief testing.
Roughly speaking, what the old code tried to do was:
Try and match entire line, if that fails, then
- Try and find patterns on the line, for each one it finds
-- Split the line up separated by " and "
--- Try and match each segment
What I changed it to was:
Try and match entire line, if that fails, then
- Split the line up separated by " and "
-- Try and match each segment
Vitality talent probably
Ah yes! Thanks, forgot about that.
For example, my Towering Mantle of the Hunt shows 123,1 RAP / 1,86% crit chance, though the actual RAP increase is only [48 + 30(Enchant) + 10,8(45% from Int with Careful Aim)] * 1,04 (2 points in Master Marksman) = 92,325, and the crit chance which should be 1,4% (21 + 10 Crit Rating) shows up as 1,86%.
I think it's counting some of the bonuses from enchants twice. I noticed the same thing yesterday when comparing my Tier 4 legs with Nethercobra Leg Armor to Tier 5 legs with no enchant.
Line:6285
replace with:
StatLogicLib is... The only variable name used all over the file is 'text', so i couldn't really get the part where suddenly 'text' becomes a table in those number of for-loops :<
Yes, it does work for every class. RatingBuster isn't a "points" mod. It simply converts combat ratings into percentages, and shows you the total effective stats for your class, factoring in talents. Although this is the kind of question that when I see, I think why don't you read the description, or just download it and see?
This that I posted in previous page is imo a better "fix" than all the tricks trying to force text to be a string.
There's a reason it's a table and the code is looping through it.
Problem was ambiguity in variable name.
Doing those changes to r33977 fixes it.
Don't know what is the current revision code, since I stopped updating after I fixed statlogic r33977.
(until Whitetooth has a chance to go over his code himself)
:) .....I have several other mods that do something like it for different class's
Could we get an option to not show the sigma graphic next to the "stat summary" header in the tooltip? I think the blank line is enough of a separator, and the icon just looks... out of place.
great addon, love it!
minor issue but no bid deal. when i click a link to an item the stat numbers appear halfway outside the frame... if i just close the window and click the same link again it shows up just fine. i don't think this works when hovering inventory items, only when i click a link to an item, ie from the chat window.
if anyone is having the same problem, any advice is most apreciated
thanks a lot
its not 112 Spelldamage its only 85
this makes it a little hard to compare items if u just rely on the compared stats and don't recheck them
How do you get those colors in you tool tip?
I could imagine that the coloring mod is adding/changing lines in the tooltip which might make ratingbuster count those lines twice.
that's the a bug in statlogiclib
whenever there is a line with 'and' then both or one of the stats are counted twice.
in your screenshot:
+22 dmg AND 14 hit
4 crit AND 5 dmg
22+5 = 27 + 85 = 112
I've got a pastey of the diff of my fix, which I believe is far less hacky than others - it reduces quite a lot of the code and seems to produce correct values from my brief testing.
Roughly speaking, what the old code tried to do was:
Try and match entire line, if that fails, then
- Try and find patterns on the line, for each one it finds
-- Split the line up separated by " and "
--- Try and match each segment
What I changed it to was:
Try and match entire line, if that fails, then
- Split the line up separated by " and "
-- Try and match each segment
http://ace.pastey.net/23732
The full modified lua file is downloadable from http://www.lim.com.au/wow/StatLogic-1.0.lua
This is a diff from r34717 of StatLogic-1.0.lua
Since I do not know StatLogic well enough, I'm not going to touch the svn version.
Could you make a post in the dev corner section, notifying that there is a bug, and you have a potential fix?
for casters/healer etc it could be very helpfull to easierly see i.e. if a 2h staff is better than the current 1h+off ;)