It's been a little while since I last posted. I've recently got back from ICFP 2008, and quite a few people asked me what I was doing now. I've also got a few comments on a a few other things. The following as a section of disjointed paragraphs on a variety of topics, both academic and personal.
Catch Talk
I gave a talk at the Haskell Symposium, about Catch. A video of the talk is now online.
Generics Talk
Alexey gave a talk about generic programming libraries at the Haskell Symposium. I was particularly interested in this talk as it is somewhat like a competition between libraries, where Uniplate is one of the competitors. One thing I noticed is that the Uniplate version of the SYB example in the talk can be written as one single lexeme, namely uniplateBi. The talk was much more about generics libraries, while Uniplate is probably more accurately described as a traversal library, so issues such as conciseness of code were left out. One thing I did disagree with from the talk was the assertion that Uniplate requires Template Haskell and Data/Typeable deriving. In reality Uniplate requires neither, but if they are present, then you have the option of using them to write even less code.
From a combination of the paper and the talk I think its fair to conclude that if Uniplate does what you want, its a pretty good choice. This fits well with the Uniplate philosophy of giving up a small amount of power, to allow a massive simplification, while still being powerful enough for most tasks.
PhD/Work
I've had my PhD viva (passed with minor corrections), and have nearly finished making the minor corrections. I'll update my website with a revised copy of the thesis shortly. I'm currently working at Credit Suisse on a three month internship. I'm not working on Haskell stuff, but instead am doing F# programming. To get a feel for some of the things that are done by Credit Suisse I recommend looking at Ganesh's ICFP talk/paper and Howard's CUFP talk. Disclaimer: Nothing I say on this blog, or anywhere public, has anything to do with Credit Suisse, but are my personal thoughts.
Personal Life
I've just moved to Cambridge, and got engaged to my girlfriend (now fiancee), Emily King. I'll be commuting to Credit Suisse for the next two months.
My Libraries/Tools
Now I'm working full-time (long hours plus a long commute), its hard for me to put the same amount of time into updating and maintaining my Haskell libraries and tools. I will still be accepting patches and answering questions, but probably not fixing too many bugs at any great speed. I'm still maintaining my bug tracker, so feel free to add bugs, fix bugs, or comment on bugs. If anyone has any particular interest in a tool, I'd consider taking on a co-maintainer to reduce some of the maintenance burden.
Well Typed
There is now a Haskell consultancy, Well Typed, comprising of Duncan Coutts and Ian Lynagh. These are two very good Haskell hackers, who are now selling their knowledge and experience. Between them, they've had substantial experience with GHC, Cabal, Hackage, ByteString, TemplateHaskell and numerous Haskell libraries. They've also taught lots of students Haskell, and helped lots of beginners on IRC and mailing lists. If I want help with Haskell, or with the general infrastructure and tools, they are usually the first people I approach. I strongly recommend that anyone needing Haskell help in a commercial environment get in contact with them - they can help you get the most out of Haskell. Disclaimer: I haven't been asked to write this section, and haven't checked with Ian/Duncan first, but I do wish them luck!
Welcome to Cambridge. Hope you like it.
ReplyDeleteCongrats on moving to Cambridge and the engagement -- both great news!
ReplyDeleteHi Neil,
ReplyDeleteCongratulations on being engaged!
Understandably your time is very short now, but if you have a moment, can you share any tips as to doing F# programming with Haskell thinking in mind? Type classes, in particular? Or can you recommend somebody else to turn to with these questions?
I've been haunted by the idea of 'porting' QuickCheck and possibly Uniplate to F#, but this pretty quickly leads to a need for type classes. And the F# type system seems too weak for them, if I am not mistaken.
So, how to survive?
Throwing type checking away, I tried to write "typeclass" instance generators based upon rules, so that this works:
everywhere<_,int> [([1;11;111], 2), ([3;33;333], 4)]
= Seq.of_list [1; 11; 111; 2; 3; 33; 333; 4]
for arbitrary nesting of tuples and lists and possibly other types you add instances of "typeclass" Everywhere to. The price is that if an instance can't be derived, you get a runtime error not a compile-time warning.
(more examples in the pastebin
http://ocaml.pastebin.com/m11ab768a)
What do you think about it? Would Uniplate be possible to port to F#, in this or some other way? Would you be interested in supervising such a project?
--Anton
Anton: I was going to blog about F# from a Haskell perspective, so I'll do that now :-)
ReplyDeleteI did actually attempt translating little bits of both QuickCheck and Uniplate to F#, and got some distance - but not hugely far. More in the blog post that will go up in a few hours!
Anton: If you just want to do some Uniplate stuff with F#, go ahead. If you want any involvement/help from me, then drop me an email - ndmitchell -ATT- gmail -DOOT- com
ReplyDelete