> By the way, it appears to me that there can be memory leaks in the GLR > parser if the stack overflows or the user invokes YYABORT, because > yyfreeGLRStack does not invoke yydestruct on the items remaining in > the parser stack. Paul Hilfinger, does this diagnosis make sense to > you?
Yes, that does seem to be true (not entirely surprising, since the destructor code that's now there isn't mine, having been added later pretty much in parallel with yacc.c). Paul H.
