Lambda the Ultimate

You may have noticed that the type systems thread dominates the Recent Posts list. :/ Wouldn't it be nice if Recent Posts only showed the last n posts in a topic, and concluded with a link like "... (click for more) ..." or something if necessary? Or even, if you were logged in, if it showed only the new (asterisked) posts in each thread?

breve is a free software package which makes it easy to build 3D simulations of decentralized systems and artificial life. Users define the behaviors of agents in a 3D world and observe how they interact. breve includes physical simulation and collision detection so you can simulate realistic creatures, and an OpenGL display engine so you can visualize your simulated worlds.

I guess the on-topic bit is this:

breve simulations are written in an easy to use language called steve. The language is object-oriented and borrows many features from languages such as C, Perl and Objective C, but even users without previous programming experience will find it easy to jump in.

but frankly steve is probably not so interesting in itself; rather breve is just a lot of fun to play with.

Apparently, several AI/robotics groups use breve for their research, but it's very easy to get started and comes with a simple IDE and several demos, of which the most interesting is undoubtedly the Walker, a genetic algorithm which evolves four-limbed walking "plates".

When a topic gets very lengthy, it can be difficult to find those items which are new. One solution is to click on the message in the chronological view, read that particular entry (only), hit the back button, chose the next, etc.

The first problem with this approach is that if I close my browser (perhaps accidentally) after reading the first item, I will lose the stars for all the rest of the unread items in the same topic, because they all show in the same page as far as the browser is concerned.

The second problem is that it requires me to constantly reload the page unnecessarily, wasting time and bandwidth.

There is a second solution, which is to search for asterisks in the text. This will quite handily find all the unread items on the page. Unfortunately, asterisks also appear in code, and *some* people *use* them for emphasis, instead of using the <em> tag. Ideally perhaps, we should just use a more distinctive search marker. In the meantime, it would be helpful if people would avoid using asterisks unnecessarily.

Thanks.

Epigram: practical programming with dependent types

Find the type error in the following Haskell expression:

if null xs then tail xs else xs

Found it? Of course you haven't. But this program is obviously nonsense! Unless you're a typechecker, that is. The trouble is that only certain computations make sense if the null xs test is True, whilst others make sense if it is False. However, as far as the type system is concerned, the type of the then branch is the type of the else branch is the type of the entire if . . . then . . . else . . ..

Statically, the test is irrelevant. Which is odd, because if the test really were irrelevant, we wouldn't bother doing it.

We mentioned this issue in discussions. LtU recently featured Epigram with emphasis on its interactive programming abilities, but I would like to add that Conor McBride's papers are mostly about practical use of dependent types for programming. They are also fun to read, though I can easily imagine how his experiments with presentation, like this one, can annoy some people.

Presentations

Besides the expected language designer presentations -

J. Gutknecht
• Criteria for Exclusion of a Construct
• Superfluous, covered by existing construct
• Problematic, contibution to problem set is
  larger than contribution to solution set
• Hidden overhead, no straightforward and
  efficient mapping to runtime exists
• Criteria for Inclusion of a Construct
  • Is extremely useful
  • Has generic character
  • Adds to completeness

and slide 9 "Complexity of syntax of programming languages" in Niklaus Wirth's presentation,

Tkachov (physicist as-end-user/programmer) provides a different perspective

and Brega's robotic art...

I've noticed that things don't seem to work as I'd expect when a thread extends onto a second page. For instance, links from the tracker don't link to the correct page. Instead it generates a link to a non-existent target in the first page, so I manually have to scroll down and go to the next page. Then, when I get there, there are no more red stars: it thinks I've already read all the comments. There may be other things, too, but those are the ones that impact me personally.

By the way, from the URLs it looks like comments-per-page is configurable, but I haven't seen a (non-URL editing) way to change that setting... Anyone?

Finally, my opinion: I'd prefer not to have threads paged at all. It's a rare thread around here that goes past 50 replies, and I'd much rather see them all on one page. Obviously if this is configurable on a per-user basis, please point me to it and I'll change my settings.

Thanks, and sorry once again for complaining. The new site really is light years ahead of the old one.

Is it generally ok to post content with character entity references (like ⇒ for &rArr;)? I mean how many people won't be able to read that?

Also, I remember seeing discussion on using "style" tag, but without any conclusion. Could we standardize some style classes by inclusion of their definitions in LtU CSS file(s)? I think we need classes for at least quoting previous comments (italic?) and for excerpts (blue italic? should we break tradition to make it different in color from default links?).

Have the settings for cookie expiration changed after the last code update? AFAICS it's now set for end of session expiration and I have to login again every time I open the browser, which is mildly annoying.

Reading A Methodology for Generating Verified Combinatorial Circuits, which involved some code generation/staging and AST optimization, I found myself wondering if there are libraries for just-in-time compilation (and then dynamic loading/linking) of simple numerical code. C seems like a decent language as a generation target (in that it's fairly assembly-like, and in its later (C99?) incarnations has fairly portable support for low level types like (and operations for) e.g. 80-bit floating point values.

Further, although I'm tempted to write everything in one language (currently C++) since it has made interoperability trivial, there must be some reasonable argument that the overhead of launching an optimizing C compiler surely gives you the leeway to say, invoke Meta-Ocaml routines that do the code-generation magic (never having embedded another language in a C++ program or called wrapped C++ in another language, I'm not sure how expensive or hard this is).

Anyone have some good references (to theory or practical implementations) or advice? Even the basic technique necessary to say, compile and dynamically load/link C code into another language (while probably somewhat platform specific) is foreign to me :)

I'm not at all committed to C as the code generation target; I just want something that compiles and optimizes loopless straight-line numerical computation well, and bonus points if it can split up the computation so that it runs in a distributed parallel fashion (N cpus per node, local memory on each node, relatively expensive network communication). It would also be great if code could be reordered to take advantage of cache locality and processor parallel-numeric abilities as well (of course this should be handled by the compiler and not addressed directly by the person generating the numerical code?). I have no idea which compilers (Intel's C compiler might be good) can do such analyses and how to encourage them to do so; by default I use gcc, which I understand is somewhat basic.

My intended application is repeated Estimation-Maximization (EM) parameter optimization; the computation for accumulating counts for each parameter as a function of the previous iteration's count can be compiled and made branchless, optimized (sharing common subexpressions, etc.), and then run a few hundred times. Thus I want to compile and load compiled code on the fly (the generated code of course depends on the input automata I'm estimating parameters for). For historical reasons, my work so far has been entirely C++, but obviously, switching to code generation/compilation would allow me to use any language as the driver as long as it can access the results of the compiled code without expensive marshalling or interprocess communication.

(apologies if this is more a compilers/tools than a language question)

IronPython is a new Python port for the CLR. The recently released version 0.6 is the first public version.