MAGIC Magic Mailing List
 
 

From: john wood (john.wood AT multigig DOT com)
Date: Thu Nov 29 2001 - 11:55:04 EST

  • Next message: R. Timothy Edwards: "reply to Calin on "magic-wrapping""

    I think there are some misunderstanding here
    and I dont want to get into any  flame wars.
    
    The magic-wrapper doesn't change the
    magic executable at all.    Magic does all the same things
    in the same way at the same speed.
    
    The  Wrapper acts like an auto-pilot onto magic 
    it automates the  'typing' of commands into magic
    when a gui button is pressed using the  TkSteal package
    and gets a look at what Magic says back to the console
    using the Tcl/Tk  Expect package
    (re. book  "Exploring Expect"  from O'reiley).
    
    A TkSteal routine forces the keystokes to be X events
    (otherwise Magic rejects them saying  'cursor not in window' 
     or something).
    
    The BLT package reparents the Magic window inside a Tk window
    and becomes managed by Tk but renders at the same speed
    as before  (Magic X routines work the same - directly to the X window).
    It works with OpenGl or X.
    
    
    GUI vs. Macro:
    
    I prefer using the macro keys on magic keys but for occasional users
    these are hard to remember.
    The keyboard also acts as a barrier for some people ever trying magic.
    
    
    Scheme vs. Python vs. Tcl/Tk:
    
    Tcl is an odd language but at least the GUI part; Tk is properly integrated with it.
    
    
    Screen sizes / Memory Footprint:
    I use a 1GHz, 512M laptop with 1400x1000 resolution on Linux
    which is luxurious compared to older Magic-machines.
    but I think Tcl/Tk is relatively lightweight by modern standards.
    
    License.
    Berkeley licence, or Artistic license.
    It's all going to be put on our web site when we get 
    www.multigig.org  active.
    
    The wrapper is a quick-and-dirty fix and will never be as good as
    a C level native port to KDE or Gnome.    I've only spent 5 days in 
    total on it, so it's not polished or anything.
    
    For us, we are developing it to make an integrated CAD evaluation
    system for our clocking and data transfer technology  (It will work a lot like a 
    clock tree insertion tool).
    
    Making the GUI independent of the application will let us port
    the GUI onto the Cadance environment as well as open source
    projects like Magic and  Electric editor.
    
    After finishing the  wrapper side of things, we are looking to  farm out
    the development of a fully C-based system.
    This will be on the free software  +  commercial support fees model.
    
    So, apart from some clock-layout algorithms and funny electomagnetic
    simulators,  the finished product will be free software.
    
    
    John
    


  •  
     
    Questions? Contact Rajit Manohar
    cornell logo