Re: RFC: Language::Logo

[zentara]

Big smile on my face! This is very cool and timely, considering the current big interest in remote robotics. All the examples ran well for me, and used surpising little cpu, well done. This is such a simple, yet elegant idea, that it will probably take a few days for me to digest all of it's possibilities. One of my first thoughts though, is to make a feedback channel, (thinking robotics here), so when you issue a command, the turtle responds back that it was a success, or if a failure, the coordinates where it was stopped. That way you could setup a artificial intelligence routine to map out accesible and non-accessible areas and paths around obstacles. It kind of reminds me of those floor sweeper mini-robots. :-)

I guess a good challenge to someone, would be to write a script to completely search an area, in the least time, with the least movement. I just saw a news report where UPS drivers are now being given delivery routes, where they only make right turns, since left turns usually mean a time consuming wait at a light, or intersection. Maybe you could use this module to map out those sort of problems, given a fixed set of coordinates to visit, a set of paths ( streets), and legal street directions?

One of the biggest early hurdles was getting past the well-known fact that you can't do Tk within a child thread

I think the Tk canvas is an excellent widget for this, but the Gnome2::Canvas, does allow for you to access Gtk widgets from within the child threads, with it's "thread-safety" mechanism. I would suggest that, but the Gnome2::Canvas development has been frozen, and has an iffy future. There is a promising replacement though, called Papyrus which you may want to look at. It dosn't have a perl port yet, but it can't be far off. All in all though, the Tk canvas is still the best thing going out there, for simple drawing.

[liverpole]

Thanks for the feedback, zentara!

    One of my first thoughts though, is to make a feedback channel

That's an excellent idea.

Actually the module already does have the server respond to the client after each command.  It did occur to me previously that it might be useful to send back a better response than just "" (blank) for success, but I never implemented it.

Let me try modifying the code to send back the client's (x, y) position (and possibly one or two other parameters); the client side can then parse the values from the response, and make them available to the calling program.

Update:  I've made some fairly large changes to the module, such that the query command (and the alias ask) are no longer necessary.  Instead, every time command (or cmd) is called, the user is passed back a hash containing all "interesting" parameters and their values.

I've got to first make sure that my tests still run, but after that I'll be posting the module to CPAN soon (hopefully this evening).

Update 2:  The code is now available at CPAN.