Thanks for the update, Carl. As one datapoint, I'm quite sure that your experience with planet has help us understand how to build a better one. Jay knows more, but from my point of view your experience and dog-food eating were a huge help. Thanks.
Robby On Mon, Sep 24, 2012 at 6:26 PM, Carl Eastlund <c...@racket-lang.org> wrote: > Short, short version: I will be working on a new Dracula implementation; see > https://github.com/carl-eastlund/dracula (currently just a bare Racket > fork). > > ==== > > Short version: While the "new Dracula" is still spiritually "ACL2 via > Racket", the new one will be very different from the old. It will be much > more Rackety, have full use of Racket macros and modules, and have a much > improved component system based very closely on ML functors. It will also > be distributed as a regular collection rather than as a planet package. > Along with Dracula development, I will "spin off" some of my support > libraries as new collections (or parts of existing ones). > > ==== > > TL;DR: > > The implementation of my thesis is "almost ready for prime-time". I want to > push it out to the world soon. I have not found Planet to be an ideal > platform for an internal PLT developer trying to keep up with the > development branch; I need to update my code far too often, and yet for > clients I have to maintain compatibility with the released version. I could > probably resolve that if I wanted to maintain "stable" and "release" > branches of my own project; I do not. Our organization has a whole > development and release infrastructure built around our core repository, and > it has not been beneficial for me to work outside that for so long. > > In the meantime, my private Dracula development has built up a number of > support libraries. Some of them are superfluous and should just basically > be "inlined" into the implementation, but some are also near-ready for > release in their own right. The top candidates are my debugging library, > which is in the spirit of unstable/debug but significantly improved, and my > pretty-printer which has a number of advantages over racket/pretty. I have > found both of these to be immensely useful in debugging programs with little > errors buried under huge amounts of data. I don't know if the unstable > collection will be part of the life cycle for these collections; I've yet to > pin down the best way to use unstable so it doesn't just turn into a code > graveyard. Probably these collections are better candidates than some of my > previous attempts, in that I am actively developing them for a specific > purpose and intend to push them out as stable collections before long. In > previous cases, I was just putting stuff "out there" and "seeing what > happens". > > The old Dracula was a Racket model of ACL2: taking ACL2 programs, and > simulating them in DrRacket so we could get the world teachpack and syntax > checking. The new Racket is an ACL2 model of Racket, designed to be much > more along the lines of taking first- (and possibly second-)order Racket > programs and certifying ACL2 models of them. I don't know how precise that > will be; the new ML-like components don't look much like anything we have > natively in Racket. > > Anyway, the repository is up on Github, linked above. Right now it's just a > copy of plt/racket, but I'll be gradually pulling pieces of my private > Dracula repo (which for various reasons I prefer to refactor into the Racket > fork piecemeal rather than copy wholesale). > > Carl Eastlund > > > _________________________ > Racket Developers list: > http://lists.racket-lang.org/dev > _________________________ Racket Developers list: http://lists.racket-lang.org/dev
