[tool] JassOptimizer

10-05-2010, 10:34 AM

Hell o, am back

thanks to conscription and the stupid bundeswehr-fuckers i had to make an involuntary pause in my mapping activities...

onTopic:
today i checked some of my maps and found lots of unperformant code, so i decided to write an optimizer for it and viola here it is.

requires microsoft .Net framework 3.5 or higher

usage:

quite simple
- get the executeable
- extract the "war3map.j" from your map or use another .j file
- execute it
- drop your beloved piece of code (.j-file)
- press enter
- wait some sec
- go to the displayed output-path and use the file with "_optimized.j"
- check the code again (if you don't trust me)
- name it "war3map.j" and place it in your map again

ATTENTION:
PLEASE FOR THE SAKE OF THE OLD GODS USE ONLY FILES W I T H O U T SYNTAX-ERRORS, otherwise the program will bug or crash

optimization-details:

if you wanna have the source, pm me...

changelog:

to do:

Attachment 49638

10-05-2010, 01:13 PM	#2
DioD	post any code before and after optimization. i suggest you to optimize code of any custom blizzard map, skibi for exs

10-05-2010, 01:57 PM	#3
Tot	attached files in first post...

10-05-2010, 04:45 PM	#4
DioD	your system mess with spaces, easy to detect on jasshelper__initstructs25167235 also constant inline failed for functions and basic mathematic operations.

10-05-2010, 06:17 PM

Tot

Quote:

Originally Posted by DioD

your system mess with spaces, easy to detect on jasshelper__initstructs25167235

can't follow you...plz explain further

Quote:

Originally Posted by DioD

also constant inline failed for functions and basic mathematic operations.

ähm yea, maybe i should mention that it does atm only inline constant GLOBALS with containing no function calls or other operations, cause in some cases the code would become slower when inlining these...
ex

JASS:

globals
constant real HardToCalc = SquareRoot(Cos(12)*Sin(32)*Atan(45)+I2R(R2I(Pow(Cos(18+Tan(8)),Atan2(6,13)))))

function Test takes real f returns nothing
local integer i=0
loop
exitwhen i>HardToCalc*f
set i=i+Atan2(HardToCalc,HardToCalc+1)
endloop
endfunction

i know it's a constructed and exaggerated example, but something like this can happen...therefore I'm searching a way to precalculate such constants...buggy atm

10-06-2010, 04:02 PM	#6
DioD	its possible to calculate this inside optimizer (just include some limits, 2-3 levels of variables) and pass actual data into inline.

10-07-2010, 08:42 AM

Tot

Quote:

Originally Posted by DioD

its possible to calculate this inside optimizer (just include some limits, 2-3 levels of variables) and pass actual data into inline.

i know, that's what i'm trying to add in the next version...ufo it currently bugs heavyly or miscalculates (2+3*4 = 20

)

what's with the space-mess?

10-07-2010, 10:39 AM	#8
DioD	your optimizer add spaces, i posted function name, check before and after. to calculate properly you shoud sort actions by priority levels or hook into some other math library.

10-07-2010, 11:35 AM

Tot

Quote:

Originally Posted by DioD

your optimizer add spaces, i posted function name, check before and after.

yea...and where's the problem/bug?

10-08-2010, 03:40 AM	#10
DioD	this is dead data, it shoud remove useless spaces not add em.

10-08-2010, 05:54 AM

#11

Tot

Quote:

Originally Posted by DioD

this is dead data, it shoud remove useless spaces not add em.

you're right

10-10-2010, 06:39 PM	#12
Troll-Brain	No offence and all, but does it more than wc3mapoptimizer ? I suppose so, but could you detail the differences ?

12-12-2010, 05:24 PM	#13
SanKakU	lol...when i want to remove empty spaces i do ctrl h and enter 4 spaces in top line, nothing in bottom line and hit all and bam, code optimized...well, at least primed for optimization. if your code adds those 4 spaces back everytime it's run...then it's trash.

03-02-2011, 10:17 AM	#14
Bribe	Having spaces in your code does nothing at all in performance nor KB size.