Well after some rewriting I'm down to this (in the optimiser): COPYING join_mv_310 index 311 with old parent 310 to index 8499 with new parent 308 //Reparent 8499 <-- 311 Gen body caller = 308, callee=join_mf_310<310> Lazy INLINING join_mf_310<310>(()) into 308 .. INPUT: join_mv_310<311> := (eq<1553>[string] (p<3791>, sep<1510>)); // begin match // match case 1:|case 1 if((lnot<2219> (match case 1)(join_mv_310<311>)))goto _ml314; return (add<3982> (p<3791>, b<3792>)); // match case 2:any _ml314: return (add<3982> ((add<3982> (p<3791>, sep<1510>)), b<3792>)); ELIMINATING VARIABLE join_mv_310<8499> -> (eq<1553>[string] (p<3791>, sep<1510>)) Removed 1 variables in 2 passes REMOVED CHILD SYMBOL index_310 REMOVED CHILD SYMBOL index_311 HIB:Examining function join_mf_300<300> for inlinable calls Renesting children of callee 306 to caller 300 -- Caller vs len = 0 -- Callee vs len = 0 Gen body caller = 300, callee=join_mc302<306> Lazy INLINING join_mc302<306>(()) into 300 .. INPUT: return (match case 1)(join_mv_300<301>); Inline tail apply : return (join_mh302<305> ()); Renesting children of callee 305 to caller 300 -- Caller vs len = 0 -- Callee vs len = 0 Gen body caller = 300, callee=join_mh302<305> Lazy INLINING join_mh302<305>(()) into 300 .. INPUT: return b<3792>; Inline tail apply : return (join_mh304<308> ()); Renesting children of callee 308 to caller 300 -- Caller vs len = 0 -- Callee vs len = 0 SYSTEM FAILURE Bug8, can't find parent of 310
The bug does NOT mean it can't find the parent of 310, it means it can't find the symbol 310. [If it could, the parent is in it] The reason is quite clear: REMOVED CHILD SYMBOL index_310 The symbol has been deleted. The question is: why is this code looking for a deleted symbol? The symbol is deleted if it has no uses, i.e. it is unreachable. Clearly it got reached. Either the usage closure is missing something (quite possible, the code in the bsym_table plaining IS wrong because it ignores parents, and a nested function implicitly uses its parent's stack frame), or, it is using stale data. Note: Erick's new code doesn't just delete a symbol: the function call deletes the symbol AND ALL CHILDREN. So if some routine had a "closure" represented as a set of all descendants, and tried to delete each one, it would crash badly, because THAT deletion would be unordered: if you deleted first a parent, then a child, well the child would already be gone. Good news: I'm onto it! :) -- john skaller skal...@users.sourceforge.net ------------------------------------------------------------------------------ Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d _______________________________________________ Felix-language mailing list Felix-language@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/felix-language
