scutigena-code — code snippets to implement algorithms in different languages

Re: [Scutigena-code] Hi scutigena devs

[<da...@de...>]

gabriele renzi <sur...@ya...> writes:

> [about ruby 1.7]

> > As an experiment, try running the Ruby ackermann
> > implementation with
> > '7' as an argument:

> > scutigena/tests/ruby/common/src] $ ruby1.7
> > ./ackermann.ruby 7

> > Does that work with 1.8 and or 1.6?

> It runs on my debian box:
> loser@irc:~$ ruby ack.rb

You are missing the argument though:

ruby1.7 ./ackermann.ruby 7

With '8' even 1.7 fails.

> but, I notice that on my win box it fails with SystemStackError. If
> that is the problem on *nix that maybe that can be fixed with
> ulimit.

We don't really know Ruby enough to fix things.  Maybe there is a
way...

> > Feel free to ask away:-) We didn't want to publicize the project
> > quite so early on, but I saw that article on slashdot and decided
> > that now was as good a time as any!

> I see.. and I found the link on osnews.com  ;)

Posted there too, for the heck of it:-)

> btw, some more questions:

> - noticed that any file in the src/ directory gets run, no matter
> the name/extension.  Is this considered as a bug or as a feature? :)

Ah... Hrm.  It could probably stand to be reduced to anything that
matches one of the available testnames.  I think I got bit once by it
running a ~ file.

> - do you think that could be possible to select just a language or a
> subset to be run ?  (a regexp as argument would be cool :)

Yes!  Definitely.  I think initially, the two args would be -tests and
-langs so that you could give it a list of tests to run for all
languages, or a list of languages to run all tests for, or only the
intersection of the specified -tests and -langs.

> - in random.python there's something strange to me:

>     gr = gen_random
>     for i in xrange(1,N):
>         gr(100.0)

> I wonder why there is that assignment. It seem to me
> that this is a name binding that could be useful when
> using python's random() as:

> gr=random.random 

> to avoid name search at any run, given that gen_random is already
> available in current module, that makes little sense to me :/

Seems odd to me too.  Does it run ok without 'gr = gen_random' ?

> (But, I understand effort is put on the framework rather than on the
> scripts, atm ;)

Well, we want to start adding tests too, but if we add a bunch and
then change the framework, it could be a lot of work to fix things up.

> Oh, and thanks for the quick answer.

You're welcome.

-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

Re: [Scutigena-code] Hi scutigena devs

[<sur...@ya...>]

 --- "David N. Welton" <da...@de...> ha
scritto: > gabriele renzi <sur...@ya...>
writes:
> 
> > Hi gurus,
> 
> Ciao!

oh, yes, I forgot to say "sorry about my english " ;)

[about ruby 1.7]
> As an experiment, try running the Ruby ackermann
> implementation with
> '7' as an argument:
> 
> scutigena/tests/ruby/common/src] $ ruby1.7
> ./ackermann.ruby 7
> 
> Does that work with 1.8 and or 1.6?

It runs on my debian box:
loser@irc:~$ ruby ack.rb
Ack(3,1): 13
loser@irc:~$ ruby -v
ruby 1.8.1 (2003-12-25) [i586-linux]
loser@irc:~$ ulimit
unlimited

but, I notice that on my win box it fails with
SystemStackError. If that is the problem on *nix that
maybe that can be fixed with ulimit.

> > Well, I think I'm going to bother you more in the
> near future, but
> > that's enough for now,
> 
> Feel free to ask away:-) We didn't want to publicize
> the project quite
> so early on, but I saw that article on slashdot and
> decided that now
> was as good a time as any!

I see.. and I found the link on osnews.com  ;)

btw, some more questions:

- noticed that any file in the src/ directory gets
run, no matter the name/extension.
Is this considered as a bug or as a feature? :)

- do you think that could be possible to select just a
language or a subset to be run ?
(a regexp as argument would be cool :) 

- in random.python there's something strange to me:

    gr = gen_random
    for i in xrange(1,N):
        gr(100.0)

I wonder why there is that assignment. It seem to me
that this is a name binding that could be useful when
using python's random() as:

gr=random.random 

to avoid name search at any run, given that gen_random
is already available in current module, that makes
little sense to me :/ 
(But, I understand effort is put on the framework
rather than on the scripts, atm ;)

Oh, and thanks for the quick answer.

=====
icq #69488917

______________________________________________________________________
Yahoo! Mail: 6MB di spazio gratuito, 30MB per i tuoi allegati, l'antivirus, il filtro Anti-spam
http://it.yahoo.com/mail_it/foot/?http://it.mail.yahoo.com/

Re: [Scutigena-code] Hi scutigena devs

[<da...@de...>]

gabriele renzi <sur...@ya...> writes:

> Hi gurus,

Ciao!

> I just subscribed this list cause, I have to admit I love
> benchmarks, no matter if they mostly mean nothing :)

Cool, glad to have you.

> BTW, I'd like to ask some questions:

> Did you choose to reuse some of the code from doug's shootout? It
> seem there are really little benchmarks (as in different tests) in
> the cvs, compared with doug's.

We have chosen to reuse some of the tests, but we held off on adding
them while we developed the framework.  If we had had too many tests,
making changes to the framework would have been difficult.  As we
begin to stabilize the framework, and generate interest in the
project, I think we can begin to add in some of Doug's tests, as well
as create new ones.

> why are you using ruby 1.7.2 ? I know you're using debian packages,
> but that was a development/unstable version of ruby, so I wonder why
> you did not choose 1.6.8 or 1.8.x (both available for debian/stable)

Salvatore added Ruby... I think the problem was that Ruby 1.8 was
having troubles with inputs that other languages could handle.

As an experiment, try running the Ruby ackermann implementation with
'7' as an argument:

scutigena/tests/ruby/common/src] $ ruby1.7 ./ackermann.ruby 7

Does that work with 1.8 and or 1.6?

Maybe Salvatore can correct me if there is another reason...

> Well, I think I'm going to bother you more in the near future, but
> that's enough for now,

Feel free to ask away:-) We didn't want to publicize the project quite
so early on, but I saw that article on slashdot and decided that now
was as good a time as any!

Saluti da Padova,
-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

[Scutigena-code] Hi scutigena devs

[<sur...@ya...>]

Hi gurus, 
I just subscribed this list cause, I have to admit I
love benchmarks, no matter if they mostly mean nothing
:)
BTW, I'd like to ask some questions: 

Did you choose to reuse some of the code from doug's
shootout? It seem there are really little benchmarks
(as in different tests) in the cvs, compared with
doug's.

why are you using ruby 1.7.2 ? I know you're using
debian packages, but that was a development/unstable
version of ruby, so I wonder why you did not choose
1.6.8 or 1.8.x (both available for debian/stable )

Well, I think I'm going to bother you more in the near
future, but that's enough for now, 
TIA

=====
icq #69488917

______________________________________________________________________
Yahoo! Mail: 6MB di spazio gratuito, 30MB per i tuoi allegati, l'antivirus, il filtro Anti-spam
http://it.yahoo.com/mail_it/foot/?http://it.mail.yahoo.com/

[Scutigena-code] mzscheme command line options changed!

[<da...@de...>]

Slightly modified hello.mzscheme:
(define (main args) (display "hello world") (newline))


VERSION 205:
@eugene [~] $ mzscheme -f hello.mzscheme -v -C hello.mzscheme 1
hello world

VERSION 203:
@ashland [~] $ mzscheme -f hello.mzscheme -v -C hello.mzscheme 1
Welcome to MzScheme version 203, Copyright (c) 1995-2002 PLT
hello world(hello.mzscheme 1)

But even worse:

203:
@ashland [~] $ mzscheme -f hello.mzscheme -v -C 1
Welcome to MzScheme version 203, Copyright (c) 1995-2002 PLT
hello world(1)

205:
@eugene [~] $ mzscheme -f hello.mzscheme -v -C 1
default-load-handler: cannot open input file: "/home/davidw/1" (No such file or directory; errno=2)

Is there any way to accomplish that on both platforms, I wonder?

Yuck:-(

-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

[Scutigena-code] "PLEAC" - possible source for tests

[<da...@de...>]

I saw this posted to the Tcl newsgroup - it looks like it might be an
interesting place to get some tests:

http://pleac.sourceforge.net/

-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

Re: [Scutigena-code] improved heapsort.tcl

[<da...@de...>]

miguel sofer <mi...@ut...> writes:

> Attached an improved heapsort.tcl (40% faster here).

Cool.

> It uses lists and [lset] instead of arrays, tries to reuse the
> result of variable assignments (readability <<), and is otherwise
> careful about avoiding some internal buglets (like && being slower
> than nested [if]s).

Ok.  I added it to the "8.4" directory.  The 'common' directory has
code that will work back to 8.3, because we wanted to be able to
support multiple versions of a language.

> Also replaced bitshift ops with multiplication - small diff, small
> runtime penalty, big improvement in readability.

Good - we want to encourage people to *read* the code from different
implementations and see what they think about it.  There is no really
objective way to evaluate readability and maintainability, but they
are obviously important factors.

Infact, I just updated the index.html page in CVS to reflect that...

Thanks!
-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

[Scutigena-code] faster lists.tcl with inlining

[miguel s. <mi...@ut...>]

Attached a 2x faster lists.tcl, inlining some function calls.

Re: [Scutigena-code] improved random.tcl

[miguel s. <mi...@ut...>]

Now with file attached ...

miguel sofer wrote:
> If we are allowed to inline code (gcc is doing it explicitly, don't know 
> about the rest), we run 3 times faster! Attached an inlining version of 
> random.tcl
> 
> It is fair to say that this is not standard tcl coding practice ... 
> except when coding for speed. tcllib has some examples.
> 
> Miguel
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: IBM Linux Tutorials.
> Become an expert in LINUX or just sharpen your skills.  Sign up for IBM's
> Free Linux Tutorials.  Learn everything from the bash shell to sys admin.
> Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click
> _______________________________________________
> Scutigena-code mailing list
> Scu...@li...
> https://lists.sourceforge.net/lists/listinfo/scutigena-code
> 

[Scutigena-code] improved random.tcl

[miguel s. <mi...@ut...>]

If we are allowed to inline code (gcc is doing it explicitly, don't know 
about the rest), we run 3 times faster! Attached an inlining version of 
random.tcl

It is fair to say that this is not standard tcl coding practice ... 
except when coding for speed. tcllib has some examples.

Miguel

[Scutigena-code] heapsort.tcl

[miguel s. <mi...@ut...>]

attached an improved heapsort.tcl (40% faster here):
  - uses lists, not arrays
  - inlines the result of assignments (readability <<)
  - avoids some buglets in the engine (like && being slower than nested)
  - replaces bitshift with arithmetic ops (readability >>)

Miguel

[Scutigena-code] improved heapsort.tcl

[miguel s. <mi...@ut...>]

Attached an improved heapsort.tcl (40% faster here).

It uses lists and [lset] instead of arrays, tries to reuse the result of 
variable assignments (readability <<), and is otherwise careful about 
avoiding some internal buglets (like && being slower than nested [if]s).

Also replaced bitshift ops with multiplication - small diff, small 
runtime penalty, big improvement in readability.

Miguel

[Scutigena-code] Idea: graphics manipulation

[<da...@de...>]

I have an idea for a fun test: manipulate an image in some way, like
shrink it or something.  Very real world, and definitely indicative of
how a language can perform certain tasks, especially if it involves
binary data.

-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/

[Scutigena-code] Welcome to the code list

[<da...@de...>]

This list is for the discussion of implementations of particular tests
For example, Ackermann's function in Ruby, Heapsort in Mzscheme and so
on.

-- 
David N. Welton
   Consulting: http://www.dedasys.com/
     Personal: http://www.dedasys.com/davidw/
Free Software: http://www.dedasys.com/freesoftware/
   Apache Tcl: http://tcl.apache.org/