- Haskell-Cafe - Haskell.org

Recommending "Design concepts in programming languages"
by Eugene Kirpichov 14 May '09

14 May '09

Hi. To all those people who are in any sense interested in PL theory I'd like to recommend the book "Design concepts in programming languages", because I am extremely impressed with it and because I, surprisingly, have not heard much about it. So, the book is about various concepts encountered in various kinds of programming languages: denotational and operational (BOS/SOS) semantics, issues of state and control, type systems, modules, modeling effects and compilation. Every concept is introduced by defining the semantics of a language that has this concept and exploring the design dimensions and issues of this concept and language. Concepts are gradually accumulated, and by the time you reach the chapter on modules you've got a CBV language with records, mutable state, polymorphic algebraic data types, a System F type system with type inference and a hint of dependent types, abstract data types and first-class dynamically loadable modules. The tools used for description are of course the good old denotational and operational semantics and typing judgements and derivation trees; but each element of those is clearly and succintly described in text; it happens to me all the time that I am reading a type reconstruction algorithm and wondering, "why does this rule have that restriction?" and it immediately turns out that in the next paragraph, the authors focus attention on why this rule has that restriction; just like if they were reading my thoughts. That's why this book feels very comfortable to me: I am absolutely sure that I won't encounter a point where I am lost and buried under the notation; but there is also not a single boring moment. I've been interested in functional programming and PL theory for 2-3 years already, and here's a brief list of the *new* things that I have learned, at least: - What do SOS and BOS mean, and why one should care, and what properties a SOS might posess (confluence and normalization, for instance) - How many features of languages can be defined in terms of simply desugaring, and how in some cases they can't - How one might use monadic style in the semantics metalanguage to greatly simplify the semantic rules for monadic concepts like state, control and error handling (the authors mention the word "monad" only once, but they use return- and bind-like operators in their semantics) - How powerful records are, and of what use are operators like "conceal" - What use is subtyping outside of OOP - How does one define CPS-style semantics and how such a style allows to add state, control and errors with minimal changes - How small yet powerful an OOP language core can be - How algebraic datatypes can be very useful in a language without static typing - How pattern matching can be desugared into CPS-style deconstructors - How many caveats are there in defining typing rules, and how a small change in them can lead to very big changes in language expressiveness - How HM type inference actually works - Why purity is important for certain polymorphism issues - What let-polymorphism means - What effect systems are - How effect reconstruction works and how it is different from type reconstruction in nature - How effect inference can prove the external purity of certain internally impure programs That's where I finished my reading for now. The remaining looks even more intriguing; for example, I don't (yet) know how functional languages are compiled and how register allocation is done. I'm afraid to sound like a salesman, but this is absolutely the best-written technical book I have ever seen in my life, and probably the most influential one for me, excluding probably SICP. Here you can check out its table of contents: http://cs.wellesley.edu/~fturbak/pubs/6821/current/contents-nov22.pdf I was impressed with the TOC even before buying the book, but turned out that I was greatly under-estimating it. Well, that's it :) -- Eugene Kirpichov Web IR developer, market.yandex.ru

8 12

Structural sharing in haskell data structures?
by Andrew Wagner 14 May '09

14 May '09

So I've been reading a lot about a (relatively) new language called Clojure. One of its goals is to make concurrency easier via a built-in home-grown STM. Anyway, one of the ways it tries to do this is to have completely immutable data structures. Every time I read a tutorial about this in Clojure, it says "...yes, it sounds awful to think that your whole data structure gets copied every time you want to make a change to it, but it's sane because of a technique called structural sharing". Yet every time I hear immutability talked about in Haskell, what I hear is "...yes, it sounds awful to think that your whole data structure gets copied every time you want to make a change to it, but it's sane because of laziness...unless you need the whole data structure...". So I'm just curious, does GHC use structural sharing or something similar? Do other implementations? Does it matter?

8 16

hsffig and duplicate typedef declarations
by Dimitry Golubovsky 14 May '09

14 May '09

Dean, I am not sure how HSFFIG would deal with duplicate typedefs (I'm afraid you get error from gcc when processing hsc2hs output). In case you haven't seen this, I put a repo for HSFFIG here recently: http://code.haskell.org/hsffig/ Before running "cabal install" run make at the toplevel of the repo first. I have fixed some glitches recently, and adjusted both hsffig and ffipkg to work with contemporary GHC and Cabal (at least 1.6.0.x). I'd be glad to get any feedback on HSFFIG. E-mail me to "golubovsky at gmail dot com". Thanks. -- Dimitry Golubovsky Anywhere on the Web

1 0

RE: [Haskell-cafe] Developing 3 dimensional interactive environments/functional objects
by paulfrancis＠friendbliss.com 14 May '09

14 May '09

1 0

PhD position in Nottingham
by vxc＠Cs.Nott.AC.UK 14 May '09

14 May '09

PhD position in Type Theory at Nottingham ----------------------------------------- A new PhD position is available in the Functional Programming Laboratory at the University of Nottingham. The topic of research for the project is "Programming and Reasoning with Infinite Structures": it consists in the theoretical study and development of software tools for coinductive types and structured corecursion. The candidate must be a UK resident with an excellent degree in Computer Science or Mathematics at MSc (preferred) or BSc level (first class or equivalent). The applicant should have a good background in mathematical logic, theoretical computer science or functional programming. (S)he should be interested doing research in type theory, constructive mathematics, category theory and foundations of formal reasoning. We offer: PhD place with living expenses (standard UK level) for 3 years. The grants also provide laptops and travel expenses for conference and workshop visits. Nottingham University provides a vibrant research environment in the Functional Programming Laboratory. Deadline for applications: 20 June 2009. Send a cover letter and your CV to Venanzio Capretta (vxc(a)cs.nott.ac.uk) Please contact me for any additional information that you need. This message has been checked for viruses but the contents of an attachment may still contain software viruses, which could damage your computer system: you are advised to perform your own checks. Email communications with the University of Nottingham may be monitored as permitted by UK legislation.

1 0

Fun with type functions
by Simon Peyton-Jones 14 May '09

14 May '09

Friends Ken, Oleg, and I have been working on a tutorial paper about type families (aka associated data types, or type functions). It's in draft at the moment, and we'd really appreciate feedback that would help us improve it. Here it is: http://haskell.org/haskellwiki/Simonpj/Talk:FunWithTypeFuns Thank you! Simon

1 0

Functional Reactive Web Application Framework?
by Petr Pudlak 14 May '09

14 May '09

Hi, I've been learning Haskell for a few months, and it has influenced my thinking about programs quite a lot. Most of my current work is creating complex web applications. Naturally, I was thinking about how to make rich internet applications (and GUI apps in general) in an (utmost :-)) functional way. Recently I found out the concepts of functional reactive programming <http://www.haskell.org/haskellwiki/Functional_Reactive_Programming>, which supersede and generalize my ideas. My question is: "Is there a FRP web application framework, or perhaps an effort to create one?" If not, I'd be willing to start such a project, but since I'm not very experienced Haskell programmer, I'd most likely need some help from somebody more experienced. If somebody would be interested, my ideas go as follows: Handling events in GUI/RIA applications can be very tedious, especially for complex applications. Instead, I'd prefer something like that: 1) Abstract what a user wants to see using a structured datatype, specifically tailored for the application. You may imagine it as something like a structured URL. This can be then viewed as a FRP behavior -- a function of the user's request depending on time. 2) The data that will be displayed on the page are a function of this structured user request and an environment, which is usually database. This is again a behavior. 3) Finally, the resulting page is a function of the data (again a behavior). 4) User actions that change the application state (update the database) could be viewed as FRP events. So instead of handling tons of events, the programmer would construct the application mostly by composing behaviors. Only state-changing user actions would be handled as events. Does anybody see these ideas as useful? Constructive criticism is welcomed. Best regards, Petr

6 7

[ANNOUNCE] Bindings for libguestfs
by Richard W.M. Jones 14 May '09

14 May '09

I added some partial bindings for libguestfs[1] here: http://git.et.redhat.com/?p=libguestfs.git;a=blob;f=haskell/Guestfs.hs;hb=H… Some very simple example programs which use these bindings: http://git.et.redhat.com/?p=libguestfs.git;a=tree;f=haskell;hb=HEAD Any comments welcome. My Haskell skills are pretty terrible, so I'm sure there are many ways these can be improved. If someone wants to look at binding the rest of the API, then please send me some patches. (Note that the Guestfs.hs file is automatically generated). BTW, I found the documentation on writing FFIs very contradictory and incomplete. For example, I was completely defeated trying to find ways to do simple stuff like passing in integers or returning booleans. *Potentially* Haskell's FFI seems like it might be one of the best out of the languages I've used so far, but it needs way more documentation and examples. Rich. [1] http://et.redhat.com/~rjones/libguestfs/ -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/

4 7

Darcs as undo/redo system?
by Peter Verswyvelen 14 May '09

14 May '09

If you work with a text editor like Microsoft Visual Studio (maybe also Eclipse, don't know), each text editor has its own undo/redo history. However, when you perform refactoring - like renaming a function - this becomes an undo/redo on multiple files together, so in a sense these changes are part of a global history. You can combine these two different kinds of mutations, and undo/redo still works as expected. I don't know how hard this is to implement, but it looks like this problem is somehow is related to Darcs theory of patches (which I don't know yet :-)

9 9

IO trouble
by Xiao-Yong Jin 14 May '09

14 May '09

Hi, I can't really describe it in the subject. So let me try to do it here. I have two functions > f :: a -> b > g :: (a -> b) -> c -> d and I use them as > gf :: c -> d > gf = g f Now I want to handle exceptions in f and redefine f as in f' > f' :: a -> IO (Either e b) So my question is how to define gf' now to use f' instead of f? > gf' :: c -> IO (Either e d) Thanks in advance. Xiao-Yong -- c/* __o/* <\ * (__ */\ <

3 3