MathGroup Archive 2009

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: How to evaluate a notebook programmatically?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg98627] Re: How to evaluate a notebook programmatically?
  • From: David Bailey <dave at removedbailey.co.uk>
  • Date: Tue, 14 Apr 2009 06:19:53 -0400 (EDT)
  • References: <grupsj$rf$1@smc.vnet.net>

Paul Ellsmore wrote:
> Hi,
> 
>  
> 
> In one notebook, "gui.nb", I have the code for a user interface. In another,
> "functional.nb" I have around 3000 lines of code, in several hundred cells.
> I want to click a button (or something similar) in gui.nb, and tell
> Mathematica to execute all the cells in functional.nb. Functional.nb will
> use values for global variables that have been assigned in gui.nb, and
> gui.nb will display the results of the execution of functional.nb. Seems
> straightforward.
> 
>  
> 
> Now in functional.nb, we group all the cells together, so that we can
> manually execute them all with a single shift-click. What is the
> programmatical equivalent of this? I have looked through all the tutorials
> etc on controlling notebooks from the front end, frontendtokens etc., but
> nothing seems to do quite what I want. The options seem to be:
> 
>  
> 
> Option 1. Merge all the cells in functional.nb, and assign that cell to a
> Function, which can then be called from gui.nb. This is highly undesirable,
> because if an error occurs, we find out about it at the end of 3000 lines of
> code, with no idea where the error occurred. The cell structure in
> Mathematica is just too valuable to throw away. I guess that using workbench
> might mitigate some of these issues, but that means my colleagues and I all
> having to learn a new (and complicated) tool.
> 
>  
> 
> Option 2. Assign each of the (hundreds of!) cells to a separate function,
> and call each of them in turn from gui.nb. A lot of hard work here with the
> potential for errors in naming and ordering the functions in the two
> notebooks, and probably an increase in execution time.
> 
>  
> 
> Option 3. Make all the cells in functional.nb initialization cells, then get
> gui.nb to open functional.nb each time we want to evaluate it, and then
> close it again. Very clunky, and again probably slow.
> 
>  
> 
> I am sure there are better options than these? Any ideas?
> 
>  
> 
> Thanks,
> 
>  
> 
> Paul.
> 
>  
> 
>  
> 
>  
> 
> Dr. Paul A. Ellsmore
> 
>  
> 
> Nanion Limited
> 
> Oxford Centre for Innovation
> 
> Mill Street
> 
> Oxford
> 
> United Kingdom
> 
> OX2 0JX
> 
>  
> 
> Tel: +44 (0) 1865 811175
> 
> Fax: +44 (0) 1865 248594
> 

If you move your code from a notebook to a .m file (often known as a 
package file, but the code does not need to be a package), you will be 
able to execute the code by calling Get.

Since version 6, .m files edit inside Mathematica beautifully - you can 
even have headings in the code (which are stored inside the .m file as 
special comments) - it is easy to forget if you are editing a notebook 
or a package file.

David Bailey
http://www.dbaileyconsultancy.co.uk


  • Prev by Date: Re: Should I be using Mathematica at all?
  • Next by Date: Assuming odd/even functions
  • Previous by thread: Re: How to evaluate a notebook programmatically?
  • Next by thread: Uninstalling Mathematica from Linux