Hacker News Reader

Comment by hythloday | Hacker News Reader

How would you implement refcounting in the presence of longjmp? I don't see a graceful way to do it.

I am not sure I grok the code, but I think something along the following lines would work:

  - a 'recovery' list of (object pointer, old refcount) pairs
  - whenever you allocate something, add (object pointer, 0) to the list
  - when you update a reference count, check whether the object is in the list.
    If it is not, add it to the list, thus remembering the original refcount
  - whenever the code decreases reference count to zero, add the object to a
   'to be deleted' list (this could possibly be a linked list chained through
    the 'refcount' fields; the old refcount will be in the recovery list)
  - when the parser reaches toplevel without longjmp, clear the first list,
    and delete any objects in the 'to be deleted' list.
  - when a longjmp occurs, walk the list and reset reference counts. For
    zero 'old refcounts', delete the object

Elegant? Not really, but with proper macros/functions, it should not be much less elegant than reference counting on its own.

That's pretty close to Deutsch-Bobrow deferred referenced counting. If you decide to implement it, you may want to read their paper about the technique.

I don't like C libraries that use longjmp for error handling. It's like killing a fly with a bazooka. There are some cases when longjmp is useful of course, but If you want proper exceptions, use a language that supports them.

The OP is doing awesome stuff and sharing it with the world, and you want to criticize an implementation detail?

Bikeshedding like this isn't useful to anyone.

I appreciate the OP's efforts, certainly sharing code is a good thing. And one of the benefits of the sharing is to get a lot of feedback. Criticizing implementation details is important to spot, discuss and hopefully fix potential problems.

But the only actionable advice you can take away from the criticism is "rewrite it in C++". I don't think that's in any way useful and I don't think it should be made.

This. The headline was a parser combinator in C. The author is working with what he's got. Can't argue with the hangups that come with it, unless he makes some claim that his language choice is somehow superior to another, which I don't see happening here.

Right - "Don't use longjmp, do X in C instead" would have been a useful comment (particularly with example code). "Don't use longjmp, use C++ instead" is much less useful.

I agree that my comment was a bit harsh, but that's not what I said. I said that if he wanted exceptions he shouldn't have used C.

As for the alternatives, well return values never killed anyone, did they? You can use gotos to simplify the control flow within the functions. Other than that, there really isn't much to explain.

I don't like longjmp because I never expect them in C. I never think "hey, the control flow might jump to some point 20 stack frames above at any moment when I enter this library". And then I use mutexes. Cue the drama. You can't even protect yourself from the stack unwinding with handler-case or try ... catch/except.

If I release code, having it criticized is a benefit. I want to learn from others, hear what they think of the design decisions, what they would have done differently, and what they think I did stupidly.

If you criticize my work, you're doing me a favor.

Reply on news.ycombinator.com