A long time ago now, I worked on an EPSRC grant that I co-authored with Andy Pitts. I was first an RA, but then I got a lectureship at Warwick, so I became a co-PI and we had the amazing good fortune of finding Claudio Russo, who took over as the RA. (Claudio and I are now colleagues at MSRC.)
The grant was on a linear lambda calculus (actually more like a linear PCF) that we dubbed Lily. We did quite a lot of work on Lily, but for one reason and another we never really published much - just a single workshop paper. (There's an A4 ring binder on the shelf above my head full of unpublished calculations of one sort or another.) Interestingly, this paper has had some impact and a number of people, including Alex Simpson and Lars Birkedal, have done some cool work following on from it.
Claudio actually implemented Lily - but we never released the code. In fact, I don't think we even documented the fact that we had implemented it!
Anyhow, I'm happy to report that Lily has risen from the ashes! Claudio and I have translated the code from Moscow ML to F#, and late yesterday afternoon, the familiar Lily REPL had returned:
Watch this space. Hopefully in a couple of months, I'll be able to explain why we're playing with Lily again. Either way - even if our experiments fail - I promise to release the bits!