> From: Leonid Leibman <http://www.raqia.com/~leonidl> > Date: Tue, 05 Jun 2001 11:17:11 -0400 > > ---Executing: html-to-ascii > > > I think (1) was explicitly disallowed since we didn't want > to parse what > > we > > have already recognized as a rule to avoid smaller thing > kicking in.... > > To enable > > it if I remember it correctly you had to add a recursive > call? > > No. It was far more difficult. And it's pretty hacky. > (That's probably > why it's slow.) The main problem is that the resulting DFA is > not a DFA > -- it's not completely deterministic. But it's good enough > for government > work. > > Why??? Meaning why not deterministic???? In my implementation, each tag says whether to go down or not. Because it's at the tag level (I thought that was the easiest to implement), the NFA to DFA converter cannot distinguish between a "go across" and "go down" tags. I suppose the "correct" implementation would not need to have this ambiguity (probably making specialized "up" and "down" tags). > He's OK. I guess there were "rolling layoffs" at his > company. (Did I already > tell you that?) He gets to take zillions of vacations, so he > enjoys his > work :-). > > I didn't get it -- was he able to keep his job? Yes, he does. (What's miraculous is that he can still get the vacation perks! Maybe it's a height thing. :-)