[Webkit-unassigned] [Bug 32722] Allow Reserved Words in Property Accessors

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=32722





--- Comment #24 from Patrick Mueller <pmuellr at yahoo.com>  2010-01-19 13:36:54 PST ---
I finally got around to profiling the parser, but didn't see anything obvious
going on.

I created a new parser-only test which was the big concat sample, only 10x
bigger (copied the text 9 times).  Saw the same sort of perf differences. 
About 1% slower.

I started looking into the generated Grammar.cpp file, comparing the sizes of
the tables; I'm left thinking that "patch so far", which created a non-terminal
named IdentOrReserved, is a better approach than the split/duplicate way it was
done in "refactored parser changes".  The parse table were bigger, the speed
difference seemed to be insigificant ("refactored parser changes" seemed to run
a bit faster, but it's hard to tell), there's by definition more code, and
there are duplicate rules to keep in sync in the .y file.

In particular, I was worried about assigning a type of <ident> to the keywords,
but this had absolutely no effect on the generated code - it's a form of
type-checking for bison, it appears.

I'm left with assuming there's no way that I can get the speed back and add the
new capability.

On the plus side, during my research, I dug up this: Bug 33860, which seems
like something interesting to try.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.