Yacc is a very important part of UNIX because it allows one to create a parser for a language with limited knowledge of parsing. Of course, if a language is very complicated, it is better to create a special purpose parser, but with yacc one can easily create parsers for most simple DSLs.
> Yacc is a very important part of UNIX because it allows one to create a parser for a language with limited knowledge of parsing.
This isn't true. You basically need to know the entire algorithm Yacc uses to use it. What does someone with limited knowledge of parsing do when Yacc tells them this?
5: reduce/reduce conflict (reduce 3, reduce 4) on c
state 5
B : x y . (3)
E : x y . (4)
I can't see the problem here. To use stdlib `rand` you need to understand that you should first `srand(seed)` and why, and that that seed needs to be unique for each program. Then, in order to use it correctly (e.g. not in a physics simulation) you should understand it's algorithm. Using programming tools require reading manual, and sometimes understanding algorithms.
This unfortunately all comes down to what you mean by "basically" and what they meant by "limited".
I think what they said is closer to the truth. I've used flex/bison a few times to create simple languages, and as long as they're parsed correctly, which is quick/easy to roughly check with a REPL (mine were in that form), conflicts can be ignored. Mine worked fine. It was a joy to use - having both flex and bison program windows open, plus bash to run the makefile, I could add a feature and test it within seconds. I'd say I have very limited knowledge of parsing, but read a few flex/bison books, enough to grasp what that message means. It's not necessary to have no conflicts, from what I understand. And it seems you have a more....professional language in mind, like in a commercial situation.
But then again, Yacc/bison isn't very important because it allows etc.... (i.e. that's not the reason it's important)
One thing is knowing about parsers, another completely different thing is knowing how to implement an efficient parser. This is the difference, e.g., between using a machine learning algorithm and implementing one from scratch.
