Reverted yesterday's change to solve the timing/perf measurement problem. Now I think the main problem was caused by the fact that redirects weren't getting inserted into the group, causing the load to terminate prematurely. That's now solved in the http protocol code.
git-svn-id: svn://10.0.0.236/trunk@41688 18797224-902f-48f8-a5cc-f745e15eee43
This commit is contained in:
@@ -2965,7 +2965,8 @@ nsWebShell::OnEndDocumentLoad(nsIDocumentLoader* loader,
|
||||
printf("nsWebShell::OnEndDocumentLoad:%p: loader=%p url=%s status=%d\n", this, loader, spec, aStatus);
|
||||
#endif
|
||||
|
||||
if (!mProcessedEndDocumentLoad) {
|
||||
if (!mProcessedEndDocumentLoad &&
|
||||
loader == mDocLoader) {
|
||||
mProcessedEndDocumentLoad = PR_TRUE;
|
||||
|
||||
if (nsnull != mScriptGlobal) {
|
||||
|
||||
Reference in New Issue
Block a user