[Shootout-list] Re: OCaml k-nucleotide

[Daniel South]

Joel Hoffman wrote:
| Are you quite sure about this? The c#, Lua, Tcl and D programs (ie,
| the successful ones) are all reporting:
|
| T 30.41
| A 30.30
| C 19.65
| G 19.63
|
| When I change my Perl program to round *down* always on the 0.005
| boundary, it completes the benchmark locally for 10000, 20000, and
| 30000. This makes no sense, but it seems to be the way things are
| working.
|
| Joel

Joel, are you using a correctly generated fasta file?
Change the output line in the Tcl program from

 puts [format "%s %0.2f" [lindex $item 0] $percent]

to

 puts "[format "%s %0.2f %0.12f" [lindex $item 0] $percent 
$percent]\t$sum\t[lindex $item 1]"

this will give you, after the usual output, percent out to 12 decimal 
places, then the sum, then the count.

When the modified Tcl program is run against your input file, do the 
sums and counts match?

Daniel South