#apache #archlinux #asterisk #centos #debian #gentoo #haskell #kde #kubuntu #lisp #math #mysql #perl #python #ruby-lang #rubyonrails #suse #ubuntu #vim #wikipedia 0 1 2 3 4 5 6 7 8 9
Top Prev 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 Next
#haskell
<lambdabot> "\"#$%&'()*+,\""
<robert_|_> hmmmm
<sorear> there's the House group - compiling Haskell to i386 - not very interesting until they qualify "Ring 0, no OS"
<sorear> @where house
<lambdabot> http://www.cse.ogi.edu/~hallgren/House/
<sorear> yhc & nhc - haskell compilers designed for extremely low runtime memory use
<robert_|_> actually the problem is that i need to compile for Linux + ARM processors
<robert_|_> but the target hardware itself is too restricted to run the compile on
<sorear> jhc - theoretically produces portable C, but IME produces mostly error messages
<sorear> GHC can target Linux/ARM
<robert_|_> is it possible to create GCC compilable code using GHC?
<sorear> -fvia-C -keep-tmp-files
<robert_|_> ah!
<hpaste> Eighty pasted "take'" at http://hpaste.org/545
<sorear> but the code is arch specific - don't try that on a PC and expect the generated C to work on ARM
<Eighty> why is that function 30 times slower than the built-in implementation of take?
<robert_|_> ok.... so I must build on an ARM
<sorear> Eighty: what is your benchmark
<Eighty> I use evaluate and rnf to calculate the result and use getCPUTime to measure the elapsed time
<sorear> @otsnack
<lambdabot> :)
<Eighty> and my test is take (10^7) [1::Int ..]
<Philippa> robert_|_: if you can get debian running on it, you're sorted - there's a package
<sorear> Eighty: Specifically, I meant "where do you get your list"
<Philippa> I need to get my zaurus running it sometime...
<sorear> Eighty: are you using -O2
<sorear> Eighty: or -O
<Eighty> i don't think so. can you use it in interactive mode?
<robert_|_> sorear: I also ready something bout unregistered compiler. Is "unregistered" code the same as processor indep. code?
<robert_|_> i.o.w.: could I (theoretically) build an unregistered i386 comp. and then create C-code, put that on the target hardware and there run a C compiler?
<sjanssen> hmm, hpaste is timing out for me
<sorear> robert_|_: not quite. it still depends on wordsize (doesn't arm use the unique wordsize of 29?) and maybe other things
<robert_|_> ok ic
<sorear> sjanssen: me too.
<robert_|_> the strange thing is, those GumStix guys gave me a C compiler to run on i386 that compiles binaries for ARM linux
<robert_|_> so that I compile on my own computer
<Eighty> sorear: adding {-# OPTIONS -O2 #-} to the top of the file didn't make it faster
<Eighty> but it might be because i'm using ghci?
<sjanssen> Eighty: are you using it in ghci?
<robert_|_> but I guess that the most promising way to go is just to ignore that, and compile on an ARM directory anyway
<robert_|_> directory = directly
<sorear> Eighty: the optimizer is disabled in ghci because it tickles bugs in the bytecode interpreter
<Eighty> okay
<sjanssen> Eighty: you need to compile the file for optimization
<nrb23> anyone used the xcode integration?
<fasta> Uhm, shouldn't the type of an expression in ghci be the same as that same expression in a file?
<sorear> Eighty: specifically, http://haskell.org/ghc/dist/current/docs/users_guide/ghci-faq.html
<lambdabot> Title: 3.10. FAQ and Things To Watch Out For, http://tinyurl.com/29dn9g
<sorear> fasta: yes, if you use all the same options
<allbery_b> hpaste.org unhappy here too
Previous Page Next Page