From: Dave Burt <dave@...>
Date: 2005-11-14T23:57:15+09:00
Subject: Re: win32ole --Test-- Ruby vs. Python

greg.rb wrote:
> I was suprised that Ruby 200% faster verses the 'static' generated
> Python com object. The Python 'dynamic' was even slower so I won't
> mention it's time.  Psyco will not help this task either, in fact it
> slows it down to 9.2 seconds.
>
> Average of 3 runs- Python time:6.16
> Average of 3 runs- Ruby time:2.89
> Ruby was 200% faster-Chunky Bacon!
> For this specific task, I am sorry to say "Python you're fired".

Wow - that's quite a surprising disparity. Also surprising is the fact you 
mentioned in the old thread but not here, that the "static" generated Ruby 
version (using olegen.rb) was slower than the straight dynamic WIN32OLE 
code.

Might there be a relationship? Might Python's time difference also be in the 
initial loading? What do you get if you loop and repeat from .Open to 
..Close, say, 20 times? (The Ruby version takes about 20 times as long to do 
that.)

And yet another surprise: Ruby out-performs the Windows Scripting Host, too, 
with this test:

Excel 1000 cells total is  25 Apr 5894
VBScript Elapsed  3.359375

(The VBScript:)

t= Timer

Set xlApp = CreateObject("Excel.Application")
xl_file = "test.xls"
xlApp.Workbooks.Open xl_file
excel_cells_total = 0.0

For r = 1 To 1000
    xlcell = xlApp.Cells(r,1).Value
    excel_cells_total = excel_cells_total + xlcell
Next

xlApp.ActiveWorkbook.Close()
xlApp.Quit

WScript.Echo "Excel 1000 cells total is ", excel_cells_total
WScript.Echo "VBScript Elapsed ", (Timer - t)


Cheers,
Dave