[Etoys] Performance improvements form 3.9 (backported) #3

[Yoshiki Ohshima]

  Marcus,

> Here's the last batch of performance changes backported from 3.9.
>
> These seem to not give too much of a performance boost, but they
> should not hurt either.

  It seems that they do give some performance improvement.  Again, I
made up two images based on #1072 image.  One is without these changes
(i.e., vanilla 1072) and another is with all changes in you #2 and #3
emails.

without:
'#1072 memory=19365744 display=1012x700x16 painter=1439.4 viewer=1693.2 scriptor=1426.0 category=2444.6 ' 
'#1072 memory=19287180 display=1012x700x16 painter=1436.4 viewer=1692.8 scriptor=1406.6 category=2442.2 ' 
'#1072 memory=19175904 display=1012x700x16 painter=1456.0 viewer=1691.2 scriptor=1443.6 category=2452.4 '

with:
'#1072 memory=19381884 display=1012x700x16 painter=1414.4 viewer=1580.0 scriptor=1346.2 category=2280.8 ' 
'#1072 memory=19299216 display=1012x700x16 painter=1414.4 viewer=1582.2 scriptor=1332.0 category=2283.8 ' 
'#1072 memory=19179744 display=1012x700x16 painter=1465.6 viewer=1587.8 scriptor=1315.2 category=2283.8 '

Again, it is over all 7% or such improvement!  It is rather exciting,
actually.

  The following is the results from the #1059 image without all of
these changes.  The improvements seems to be "linear".  It is really
good.

without:
'#1059 memory=19310712 display=1012x700x16 painter=1483.2 viewer=1802.4 scriptor=1487.2 category=2628.8 ' 
'#1059 memory=19228056 display=1012x700x16 painter=1478.4 viewer=1796.8 scriptor=1470.4 category=2625.2 ' 
'#1059 memory=19022612 display=1012x700x16 painter=1565.2 viewer=1803.8 scriptor=1468.4 category=2627.0 '

> "Change Set:		RemoveFlag-md
> Date:			16 October 2006
> Author:			Marcus Denker
> 
> This just removed a
> 	self flag: #bob

  Oh, boy.  Even this wouldn't give us any improvement, I love this
one.

-- Yoshiki