[clean-list] Languages Comparison

[John van Groningen]

Brent Fulgham wrote:
>
>>Hash Access (hash):
>>===================
>> 
>This causes a compile error:
>cp /opt/shootout/shootout/bench/hash/hash.clean hash.icl
>/usr/bin/clm -b -nt -ou -l /usr/lib/clean/lib/ArgEnvUnix/ArgEnvC.o -I ../Include/clean -I ../../Include/clean hash -o hash.clean_run
>Type error [hash.icl,84,htNew]:"argument 2 of HashTable" cannot unify {![!Item v6!]} with v10 [v9]
>Compiling hash
>make[1]: *** [hash.clean_run] Error 1
>"(echo "BUILD COMMANDS FOR: hash.clean" ; echo ; date ; echo ; make --no-print-directory -f ../../Makefile.mb hash.clean_run) > clean_runlog.make.tmp 2>&1" Failed
>make: *** [plot] Error 1
>
>Can you look at the posted source (http://aliotto see what's wrong?

I forgot to change [] into [|] in htNew:

htNew n = { nBuckets = nprime
        , table = {[|] \\ i <- [0..nprime-1]}
        }

>>Spell Checker (spellcheck):
>>===========================
>> 
>I've done something wrong here -- I get no output.

I hadn't noticed the second call to fend in this program. I changed
missingWords to:

missingWords f ht missing	
	# (line, f) = freadline f
	| size line==0 = map withNewline missing
	| (htHasKey line ht) = missingWords f ht missing
	= missingWords f ht [line:missing]	
	where
	withNewline s
		= if (s.[size s - 1] == '\n') s (s+++"\n")

And the indentation of readWords should have been:

    readWords f ht
        # (line, f) = freadline f
        | size line==0
            = ht
            = readWords f (htAdd line 1 ht)

>>Statistical Moments (moments):
>>==============================
>> 
>There's a slight error here:
>
># Benchmarking: moments.clean
># Running: moments.clean_run -h 10000000 1
>L1: median: 250.000000
>
>L2: median: 250.500000

I had assumed that e in quicksort was the last index+1, but it is
the last index, so +1 has to be added in the test in quick_sort0:

        quick_sort0 b e k a
            | k>=b && k<=e+1

Because reading reals in Clean was still a lot slower than in c, I have
updated the runtime system patch at:

http://www.cs.kun.nl/~clean/download/Clean21/linux/patch_io_redirection.tgz

It now uses atof instead of sscanf to convert the string to a real. The
performance of the other programs in the shootout does not change compared
to the previous patch.

And a runtime system with faster stdio redirection is now also available for
windows:

http://www.cs.kun.nl/~clean/download/Clean21/windows/patch_io_redirection.7z


>>Word Frequency (wordfreq)
>>=========================
>> 
>For some reason, this compiles fine, but I get this runtime error:
>
>Run time error, rule 'ht_buckets_to_list;23' in module 'wordfreq' does not match

The indentation of the last alternative of ht_buckets_to_list was
changed by accident when I replaced tabs by spaces, it should be:

        ht_buckets_to_list :: *[!Item a!] [Item a] -> [Item a]
        ht_buckets_to_list [|i:is] l
            = ht_buckets_to_list is [i:l]
        ht_buckets_to_list [|] l
             = l

Regards,

John van Groningen