because we are now using a parser for the Editor, it's crucial that every small test pr token, is done the outmost optimal way! Any simple small mistake has an impact of 1-2 ms pr parse, which in the end mean that the Editor feel slow and laggy. 
most tokens are taken by the large Switch in jsParser - but un-cased tokens has to be checked manually in the default: - therefore i've just setup a small benchmark of what method is fastest in which 'environment'...
4x charAt()
var P = 100000
var s = "//pingeling"
var y = 0;
var d = new Date();
do{
if( s.charAt(0)=="/" && s.charAt(1)=="/")
y++;
if( s.charAt(0)=="/" && s.charAt(1)=="/")
y++;
}while(--P)
console.log( (new Date()).getTime() - d.getTime() )
2x substr()
var P = 100000
var y = 0;
d =
new Date();
do{
if( s.substr(0,2)=="
//" )
y++;
if( s.substr(0,2)=="
//" )
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
UPDATE! - 2x indexOf()
var P = 100000
var y = 0;
d =
new Date();
do{
if( s.indexOf("
//")==0 )
y++;
if( s.indexOf("
//")==0 )
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
2x search()
var P = 100000
var y = 0;
d =
new Date();
do{
if( s.search("
//" )>=0)
y++;
if( s.search("
//" )>=0)
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
2x match()
var P = 100000
var y = 0;
d =
new Date();
do{
if( s.match(/^\/\
// ))
y++;
if( s.match(/^\/\
// ))
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
4x mixed(), 1 match
var P = 100000
var y = 0;
var d =
new Date();
do{
if( s.charAt(0)=="*")
y++;
if( s.charAt(0)=="?")
y++;
if( s.substr(0,2)=="/*" )
y++;
if( s.substr(0,2)=="
//" )
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
4x mixed(), 1 match, 1 pre-match variable
var P = 100000
var y = 0;
var d =
new Date();
do{
var sub = s.substr(0,2)
if( s.charAt(0)=="*")
y++;
if( s.charAt(0)=="?")
y++;
if( sub=="/*" )
y++;
if( sub=="
//" )
y++;
}
while(--P)
console.log( (
new Date()).getTime() - d.getTime() )
The results:
4x charAt()
|
1110 ~ 1125ms |
2x substr()
|
750 ~ 797ms
|
2x indexOf()
|
640 ~ 650ms
|
2x search()
|
1125 ~ 1172ms
|
| 2x match() |
766 ~ 828
|
| 4x mixed(), 1 match |
953 ~ 969
|
| 4x mixed(), 1 match, 1 pre-match variable |
984 ~ 984
|
UPDATE hours later
Earlier i conclided that substrings was the fastest... but using indexOf(..)==0 is blazing faster then either of all ! even up to 2x faster...
real world result : a difference from 9ms to 6ms pr render, for the Editor !! (update: same speed, but in some flooded cases, it will prove better)
I think ill re-make this test...
indexOf is only faster when the test subject always return true... in a mixed environment, i think substring is faster afterall...