algorithm by neil, function per Seno.Aiko@gmail.com, whatever's left of the patch is by me :)
r=neil sr=neil
git-svn-id: svn://10.0.0.236/trunk@211878 18797224-902f-48f8-a5cc-f745e15eee43
r=bz sr=bz
Perhaps an explanation. String concatenation is a chance for garbage collection.
If you have unbounded input, this can be pretty bad. When you start running low
on memory, the garbage collector gets desperate and won't take shortcuts. Worst of
all, the original code generates one string of garbage every iteration starting with
the second or third.
This code uses a=[] to construct an array and a[n] to set the upper array boundary,
0 being the lower, and there are n slots between a[0] and a[n].
join() will build a string with n copies of the join argument, where n is the number of slots between between cells. So for:
a=[]; a[1]=''; there's 1 slot between [0] and [1]. The array's length is actually 2,
but the joined length will be 1.
While it is possible to use new Array(x), the behavior is different because the
array has length x and there are x-1 spaces between cells. This leads to code which
would be both much longer and harder to read. And definitely not particularly
intuitive.
The code used here otoh, while magical will hopefully require you to pause, think,
and quickly understand what's going on. If not, you paused long enough for this
bubble to appear and explain it to you :).
git-svn-id: svn://10.0.0.236/trunk@207376 18797224-902f-48f8-a5cc-f745e15eee43
DOMNodeList goodness that all nodelists should have. Change some JS to not
break, and some other JS to be a little more efficient with the new world of
lazy listness. Bug 240186, r=neil on the JS changes, r=jst on the content
changes, sr=jst
git-svn-id: svn://10.0.0.236/trunk@154907 18797224-902f-48f8-a5cc-f745e15eee43
