Lambda the Ultimate

I thought that Dylan was somehow revived with the donation of Functional Developer to the gwydion dylan group.
Sure, they're a small group, but there seem to be a bit of activity still.

Yup, Gwydion Dylan is active. And it's great that the functional developer stuff has been donated.

There are also people that have the complete Critical Mass's Modula-3 codebase, and have permission to release it. Hasn't happened yet though. Many of the Digital SRC applications would be prime targets for archeology and restoration.

While it is true that Agda I being actively worked on it is not a programming language. It is a logical framework for doing mathematics. You cannot run your Agda programs in any practical sense of the word. So I would say that Cayenne really is dead. Which is a bit sad since I really liked it. Especially its module system.

But I've tried to push various people to revive Cayenne or some variant of it. They are generally positive about the idea but no one wants to do the dirty work.

There is some additional information on Timber at Luleå University of Technology. The interpreter (ti) is still available from OGI but it has not received much attention in the last couple of years.

The compiler (timberc) is currently in a state which makes it unusable for third parties. I hope that there will be some documentation, some examples and a useful compiler before 2006.

Reading the articles listed on those pages should give a pretty good insight about many of the ideas behind Timber.

As for what O'Haskell does better or worse than Timber I can not answer, if someone is interested I could point Johan Nordlander to this place and he could answer once he gets back from his vacation.

Some of the differences between the two are that Timber is not lazy and it is possible to do WCET analysis for Timber programs.

On the implementation side Martin Kero has implemented a garbage collector as a part of his master thesis and that should appear in the CVS rather soon. Another implementation difference is that one should not need to write glue code in the run time system to make it work with for example TK, like one had to with O'Haskell.

So which new Haskell derivative offers the same features as O'Haskell

Reactive Objects for Haskell by Stefan Heimann and Matthias Neubauer http://www.informatik.uni-freiburg.de/~heimann/rhaskell/rhaskell-tfp04.pdf

From the abstract

Only a few programming languages exist that make the concept of reactive objects available to programmers. Unfortunately, all those systems seem to be either not actively maintained anymore or not really matured yet. In this paper we demonstrate how reactive objects can be incorporated into the existing general purpose programming language Haskell by making use of common extensions to Haskell. Our implementation is heavily inspired by the O'Haskell programming language. While O'Haskell is a standalone programming language not compatible with Haskell, our implementation comes as a Haskell library that smoothly integrates the concept of reactive objects into the world of regular functional programming.

Your perception that Timber is not maintained is fortunately wrong. The interpreter hasn't gotten much attention in the last years, however the compiler is actively being worked on. There is also a technical report and two articles written about Timber in the last couple of years.

Is there a publicly available Timber compiler?

The compiler is available via anonymous CVS as described here. As stated in an earlier comment, its current state makes it pretty unusable though.