Going to start this one as RDF .. I'm sure it will migrate elsewhere, maybe aurora/sidebar? Sorry. Something calls nsGlobalWindow_RunTimeout .. and it calls into JavaScript, which calls into XUL/RDF and then stuff leaks. A lot of different things. Several Purify stacks: MLK: 924 bytes leaked in 21 blocks * This memory was allocated from: malloc [rtlib.o] __bUiLtIn_nEw [libgcc.a] __builtin_new [rtlib.o] NS_NewXULAttribute(nsXULAttribute**,nsIContent*,int,nsIAtom*,const nsString&) [nsXULAttributes.cpp:88] RDFElementImpl::SetAttribute(int,nsIAtom*,const nsString&,int) [nsRDFElement.cpp:1815] RDFXULBuilderImpl::AddAttribute(nsIContent*,nsIRDFResource*,nsIRDFNode*) [nsRDFXULBuilder.cpp:2254] RDFXULBuilderImpl::OnAssert(nsIRDFResource*,nsIRDFResource*,nsIRDFNode*) [nsRDFXULBuilder.cpp:874] CompositeDataSourceImpl::OnAssert(nsIRDFResource*,nsIRDFResource*,nsIRDFNode*) [nsCompositeDataSource.cpp:1167] InMemoryDataSource::Assert(nsIRDFResource*,nsIRDFResource*,nsIRDFNode*,int) [nsInMemoryDataSource.cpp:1047] CompositeDataSourceImpl::Assert(nsIRDFResource*,nsIRDFResource*,nsIRDFNode*,int) [nsCompositeDataSource.cpp:796] RDFXULBuilderImpl::OnSetAttribute(nsIDOMElement*,const nsString&,const nsString&) [nsRDFXULBuilder.cpp:1410] XULDocumentImpl::OnSetAttribute(nsIDOMElement*,const nsString&,const nsString&) [nsXULDocument.cpp:3720] RDFElementImpl::SetAttribute(const nsString&,const nsString&) [nsRDFElement.cpp:889] ElementSetAttribute(JSContext*,JSObject*,unsigned int,long*,long*) [nsJSElement.cpp:260] js_Invoke [jsinterp.c:650] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Execute [jsinterp.c:815] JS_EvaluateUCScriptForPrincipals [jsapi.c:2390] JS_EvaluateUCScript [jsapi.c:2371] JS_EvaluateScript [jsapi.c:2339] GlobalWindowImpl::RunTimeout(nsTimeoutImpl*) [nsGlobalWindow.cpp:1180] nsGlobalWindow_RunTimeout(nsITimer*,void*) [nsGlobalWindow.cpp:1101] * Block of 44 bytes (21 times); last block at 0x11df530 MLK: 666 bytes leaked in 28 blocks * This memory was allocated from: malloc [rtlib.o] JS_malloc [jsapi.c:730] js_InflateString [jsstr.c:2291] js_Atomize [jsatom.c:520] JS_InternString [jsapi.c:2539] nsXPCWrappedNativeClass::BuildMemberDescriptors() [xpcwrappednativeclass.cpp:149] nsXPCWrappedNativeClass::nsXPCWrappedNativeClass(XPCContext*,const nsID&,nsIInterfaceInfo*) [xpcwrappednativeclass.cpp:93] nsXPCWrappedNativeClass::GetNewOrUsedClass(XPCContext*,const nsID&) [xpcwrappednativeclass.cpp:66] nsXPCWrappedNative::GetNewOrUsedWrapper(XPCContext*,nsISupports*,const nsID&) [xpcwrappednative.cpp:119] XPCConvert::NativeData2JS(JSContext*,long*,const void*,const nsXPTType&,const nsID*,unsigned int*) [xpcconvert.cpp:269] nsXPCWrappedNativeClass::CallWrappedMethod(JSContext*,nsXPCWrappedNative*,const XPCNativeMemberDescriptor*,int,unsigned int,long*,long*) [xpcwrappednativeclass.cpp:586] WrappedNative_CallMethod(JSContext*,JSObject*,unsigned int,long*,long*) [xpcwrappednativeclass.cpp:665] js_Invoke [jsinterp.c:650] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Execute [jsinterp.c:815] JS_EvaluateUCScriptForPrincipals [jsapi.c:2390] JS_EvaluateUCScript [jsapi.c:2371] JS_EvaluateScript [jsapi.c:2339] GlobalWindowImpl::RunTimeout(nsTimeoutImpl*) [nsGlobalWindow.cpp:1180] nsGlobalWindow_RunTimeout(nsITimer*,void*) [nsGlobalWindow.cpp:1101] TimerImpl::FireTimeout() [nsTimer.cpp:73] nsTimerExpired [nsTimer.cpp:189] g_timeout_dispatch [gmain.c:1147] g_main_dispatch [gmain.c:647] g_main_iterate [gmain.c:854] g_main_run [gmain.c:912] * Block of 48 bytes at 0x14f1080 * Block of 42 bytes at 0x14d1e60 * Block of 32 bytes (2 times); last block at 0x1187ac8 * Block of 30 bytes (3 times); last block at 0x1182b70 * Block of 28 bytes (2 times); last block at 0x1184a50 * Block of 26 bytes (3 times); last block at 0x1183958 * Block of 24 bytes (2 times); last block at 0x1179038 * Block of 22 bytes (3 times); last block at 0x117de40 * Block of 20 bytes (2 times); last block at 0x13e5a80 * Block of 18 bytes (3 times); last block at 0x147f118 * Block of 16 bytes (2 times); last block at 0x11f4408 * Block of 14 bytes (2 times); last block at 0x149aa88 * Block of 12 bytes at 0x118d528 * Block of 8 bytes at 0xea4690 MLK: 571 bytes leaked at 0x13d85e0 * This memory was allocated from: malloc [rtlib.o] PR_Malloc [prmem.c:38] NET_BACat [nsNetStubs.cpp:2053] net_ChunkedWrite [cvchunk.c:59] net_pull_http_data [mkhttp.c:3170] net_ProcessHTTP [mkhttp.c:3570] NET_ProcessNet [mkgeturl.c:3355] NET_PollSockets [mkselect.c:320] nsBlockingStream::Read(char*,unsigned int,unsigned int*) [nsNetStream.cpp:711] rdf_BlockingParse(nsIURL*,nsIStreamListener*) [nsRDFXMLDataSource.cpp:482] RDFXMLDataSourceImpl::Open(int) [nsRDFXMLDataSource.cpp:724] ServiceImpl::GetDataSource(const char*,nsIRDFDataSource**) [nsRDFService.cpp:923] *unknown func* [pc=0xef6e321c] nsXPCWrappedNativeClass::CallWrappedMethod(JSContext*,nsXPCWrappedNative*,const XPCNativeMemberDescriptor*,int,unsigned int,long*,long*) [xpcwrappednativeclass.cpp:540] WrappedNative_CallMethod(JSContext*,JSObject*,unsigned int,long*,long*) [xpcwrappednativeclass.cpp:665] js_Invoke [jsinterp.c:650] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Invoke [jsinterp.c:666] js_Interpret [jsinterp.c:2199] js_Execute [jsinterp.c:815] JS_EvaluateUCScriptForPrincipals [jsapi.c:2390] JS_EvaluateUCScript [jsapi.c:2371] JS_EvaluateScript [jsapi.c:2339] GlobalWindowImpl::RunTimeout(nsTimeoutImpl*) [nsGlobalWindow.cpp:1180] nsGlobalWindow_RunTimeout(nsITimer*,void*) [nsGlobalWindow.cpp:1101] TimerImpl::FireTimeout() [nsTimer.cpp:73] nsTimerExpired [nsTimer.cpp:189] g_timeout_dispatch [gmain.c:1147] That's enough to get the idea across for now.

Bruce, I've cleaned up a number of mlks over the last couple of days (last one went in this morning, 19990615). I'm going to mark it "fixed" for now: if you find it's still there, just re-open the bug.Thanks.

verifying fix for now. if it's still there, you know the drill...