[Webkit-unassigned] [Bug 20574] bindings need better internal support for binding-object-to-webkit-c++-object mapping

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=20574





------- Comment #2 from lkcl at lkcl.net  2008-08-31 06:51 PDT -------
have you considered doing a test evaluation - especially on AJAX web sites -
to get some statistics on how many entries there are in the various HashMaps?

bearing in mind that the HashMaps are... at least double the amount of memory
of a single per-object pointer, then if over 50% of the c++ DOM model objects
have javascript binding objects attached to them, you're not saving any memory
at all, and adding complexity andd overhead to boot.

if the overhead of the HashMaps are _three_ times that of a single per-object
pointer, then ...

so, i believe it may be worthwhile adding in some printfs into addWrapper in
JSDOMBindings.cpp to print out the count of the number of HashMap entries, and
also have a count printout of the number of c++ DOM objects.

even if it turns out that, even on heavy-javascript-based web sites such as
gmail, yahoo mail and other sites, that the memory overhead of the HashMaps is
significantly less than having a single pointer in the c++ DOM objects which
goes unused (perhaps a lot of the time), the issue of my having to
cut-and-paste and global-search-replace a good 1,500 line of code to get a
corresponding object map for the glib bindinds remains.


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