From: Ralf <lausianne@...>
Date: 2003-02-03T18:33:26+09:00
Subject: Re: tkgnuplot problem: fork() on win (popen3 for windows?)

Here you go:

Code:
require 'gnuplot.rb' # maybe need change the path of library
gp = GnuplotProcess.new('c:\program 
files\design\gnuplot\gp371w32\pgnuplot.exe')
gp.cmd('plot sin(x)')
sleep 5

Result:
C:/Program Files/ruby/lib/ruby/1.6/gnuplot.rb:105:in `write': Invalid 
argument (Errno::EINVAL)
	from C:/Program Files/ruby/lib/ruby/1.6/gnuplot.rb:105:in `_send_cmd'
	from C:/Program Files/ruby/lib/ruby/1.6/gnuplot.rb:96:in `synchronize'
	from C:/Program Files/ruby/lib/ruby/1.6/gnuplot.rb:96:in `_send_cmd'
	from C:/Program Files/ruby/lib/ruby/1.6/gnuplot.rb:154:in `cmd'
	from C:\DOCUME~1\longwitz\LOCALS~1\Temp\rb13C.tmp:3
Completed(1)


nagai@ai.kyutech.ac.jp wrote:
> Hi, 
> 
> From: Ralf <lausianne@gmx.net>
> Subject: Re: tkgnuplot problem: fork() on win (popen3 for windows?)
> Date: Fri, 31 Jan 2003 23:25:17 +0900
> Message-ID: <3e3a3587$1@epflnews.epfl.ch>
> 
>>nagai@ai.kyutech.ac.jp wrote:
>>
>>>Well, what happen when you give the gnuplot command path to 1st argument of
>>>GnuplotProcess.new method ? 
>>
>>same thing.
>>Then I found this in our script:
>>     gp.terminal = 'x11 1'
>>Smells like Unix to me ... well, commenting it out didn't help ...
> 
> 
> When only require it, the part which include "gp.terminal = 'x11 1'" 
> must not be evaluated. 
> Did you execute "ruby gnuplot.rb"?
> I want to know the result with all messages 
> of running the following script.
> -----<gp-test.rb>-------------------------------------------------------
> require './gnuplot.rb' # maybe need change the path of library
> gp = GnuplotProcess.new('c:\program files\design\gnuplot\gp371w32\pgnuplot.exe')
> gp.cmd('plot sin(x)')
> slesp 5
> ------------------------------------------------------------------------
> If you can check the running process list, 
> please check the exsistence of gnuplot process just after executing 
> the second line of the script. 
> The information of "When is the gnuplot process lost?"
> gives me hints of the problem.
> 
> # If the problem depends on win32_popen, 
> # I cannot fix the problem immediately. 
> # Because I must finish to write two papers in this five days. :-)
> --
>                                   Hidetoshi NAGAI (nagai@ai.kyutech.ac.jp)
>