On Mon, 2012-01-30 at 10:15 +0000, lkcl luke wrote: > On Mon, Jan 30, 2012 at 10:03 AM, Kees Bos <[email protected]> wrote: > > On Mon, 2012-01-30 at 09:41 +0000, lkcl luke wrote: > >> we kinda threw stuff in to see what worked. at one point, the use of > >> x['abcd'] instead of x.abcd was added - this unfortunately has a > >> rather large performance hit on IE's trident engine, so at some point > >> should go back perhaps to x.$abcd. > > > > The use of x['abcd'] needs less attribute mapping that x.abcd (esp. in > > IE if I remember correctly). I think it was also introduced before we > > implemented the attribute mapping. And there's a second thing (I think): > > the google code compressor doesn't mess with strings, but does shorten > > attribute names. > > > > But the major problem with x.$abcd (note the $) is: > > > > DOM.py: > > ... > > def getOffsetHeight(elem): > > return elem.offsetHeight > > ... > > > > this (c/w)ould be translated to: > > function $getOffsetHeight($elem) { > > return $elem.$offsetHeight; > > } > > > > and the problem is $offsetHeight. > > mrrhhmmm... good point. that would get compressed, which would > mangle the whole thing.
Yep, but that's not my major point :-) offsetHeight is a dom element attribute and must be addressed as $elem.offsetHeight and not as $elem.$offsetHeight. So, in this case the translation from py x.foo to x.$foo would be erroneous. Could be catched with something like (typeof $x.foo == 'undefined'? $x.$foo : $x.foo)
