[antlr-interest] Serious Bug when using BitSetgeneration

[Bryan Ewbank]

Let's back up a bit here, please...

We all agree that bugs are bad.  This BitSet stuff is a potential bug.
I've changed my production 2.x ANTLR code to avoid them where possible,
and I'm sure others have as well.

I would encourage you and others that do have time, or other motivation,
to continue exploring the bug and figuring out exactly what's happening,
and when.

A few months ago I would have eagerly attacked this once a good
reproduceable testcase was available.  Now, I cannot.  It's that simple.

I also think that we all understand that ANTLR is open source, and that
people have other things calling them.

What we should not expect is that there are people available to help at
the drop of a hat; there may be some, certainly, but there may not at a
particular time, either.  That's a risk.

On 11/18/05, Geir Ove Skjaervik <geiroves at online.no> wrote:
> c) The workaround means avoiding BitSets alltogehter for ** everyone **.
> As long as we (programmers and the ANTLR developemnt team) does NOT
> understand the nature of the problem, then we do NOT know when this bug
> strikes!
>
> d) Avoiding BitSets all the time means two things: i) Performance takes
> a big hit ii) Why are they there if all they can do is pose a BIG
> problem in unknown situations