Summary: | JSC needs a NodeWalker class to walk the AST tree | ||
---|---|---|---|
Product: | WebKit | Reporter: | Eric Seidel (no email) <eric> |
Component: | JavaScriptCore | Assignee: | Nobody <webkit-unassigned> |
Status: | RESOLVED INVALID | ||
Severity: | Normal | CC: | barraclough, ggaren, sam |
Priority: | P2 | ||
Version: | 528+ (Nightly build) | ||
Hardware: | Mac | ||
OS: | OS X 10.4 | ||
Bug Depends on: | |||
Bug Blocks: | 16547 |
Description
Eric Seidel (no email)
2007-12-21 22:02:52 PST
As a further optimization, we could actually let the beginCallback return some sort of information as to if the endCallback is needed or not. Still pondering... I'm not convinced that an abstract NodeWalker is the best direction right now, for two reasons: 1. Making tree traversal specific to the operation at hand (e.g., declaration gathering doesn't visit as many nodes as variable access optimization) was a substantial performance win, at least when it was first implemented. 2. The upcoming move to byte code will substantially change our AST needs in unpredictable ways. The big win is that it's non-recursive. Meaning when we move to byte-code, we can support much much deeper AST trees. Adding ggaren so he'll see my comment. AST traversal is already non-recursive. yes, you're right. I'm OK leaving the NodeWalker patch in bug 16547 out of the tree. I don't think this bug is still relevant. |