This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Open boost project; start opening files one by one. On 5-th - 15th file, error highlighting does not appear. Stack dump shows stack overflow.
Created attachment 70650 [details] stack
Created attachment 70653 [details] another stack
It seems it's rather a very deep stack that causes it to be very slow. At leas I haven't seen stack overflow exception.
File: subrule.opp Construct: template <int N, typename ListT> struct get_subrule { typedef typename get_subrule<N, typename ListT::rest_t>::type type; }; For typedef type, the chain of original types seems to go forever
Please do not forget to evaluate IZs when you look at them.
Here is a localized code example template<typename T1, typename T2> class subrule_list { }; template<int T1, typename T2, typename T3> class subrule_parser { }; template <int N, typename ListT> struct get_subrule { typedef typename get_subrule<N, typename ListT::rest_t>::type type; }; template <typename ParserT, typename ScannerT> struct parser_result { //typedef typename parser_type::template result<ScannerT>::type type; }; template <int ID, typename ScannerT, typename ContextResultT> struct get_subrule_result { typedef typename get_subrule<ID, typename ScannerT::list_t>::type parser_t; typedef typename parser_result<parser_t, ScannerT>::type def_result_t; };
After that happens, IDE is still functional; only highlighting feature does not work any more.
I see the issue is in o.n.m.c.modelimpl.csm.Instantiation.getClassifier: it tries to instantiate the element that is already an instantiation; the newly created instantiation (that wraps a previously created one)
Created attachment 70783 [details] Proposed fix
Please disregard the attached patch ("proposed fix"). The proper fix is pushed into the CND integration repository. Fix: http://elif.russia.sun.com/hg/release65_cnd_freeze/rev/144890364e4d Test: http://elif.russia.sun.com/hg/release65_cnd_freeze/rev/1954070e8dce QA, verify please.
Verified. Inaccuracy tests works without problem on Boost.
Fixed in main fix: http://hg.netbeans.org/main/rev/5245b8fd6edd test: http://hg.netbeans.org/main/rev/2b3c35a070ca
Fix is verified, no objections. Just check that performance did not regress.
I measured initial boost parsing; I see no regression.
Visiting all references on ddd project does not show regressions either.
Integrated into 'main-golden', will be available in build *200810011401* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main/rev/5245b8fd6edd User: Vladimir Kvashin <vkvashin@netbeans.org> Log: Fix for #148453 Highlighting thread hangs on boost