From: "D.Hering" <v.... gmail.com> wrote: > On May 23, 4:04 am, Tim Golden <[EMAIL PROTECTED]> wrote: > > Robert Rawlins - Think Blue wrote: > >
I did not see the original post either :-( ... > > > I've got an application that runs on an embedded system, the application > > > uses a whole bunch or dicts and other data types to store state and other > > > important information. > > > I'm looking to build a small network of these embedded systems, and I'd love > > > to have them all share the same set or data. Is it possible to share the > > > applications variables across multiple applications, so certain lists are > > > like a 'pool' written to by the different systems? I'm sure I could cobble > > > something together by writing the lists to shared files instead of keeping > > > them in RAM, but that feels a little inefficient. I'd like to try and > > > configure some form of master/slave relationship between my applications if > > > possible. What kind of "embedded" platform are you using? - sounds like a cut down PC. I have just spent a lot of time solving similar problems on *very* small processors, and ended up designing a very simple language and a kind of distributed OS, that can do nothing more than putting voltages on wires, but both locally and remotely... The human interface part runs on PC's and is written in Python. The embedded controllers are written in a mix of C and assembler and implement an assembler like reverse polish notation multi tasking language running on a virtual risc machine. The "compiler" is also written in Python. Can probably only share some of the pain with you to make your journey less painful... > > > > I was really surprised you hadn't received a whole > > slew of answers for this (even if they were: search > > the newsgroup for the last time this was asked!) But > > then I noticed that the post hadn't appeared on Google > > Groups, at least. I read things via the mailing list; > > is it possible your post hasn't made it across to > > Usenet either? > > > > Just to get the ball rolling, I'd suggest two things: > > > > Pyro -http://pyro.sf.net This is good advice, if you have the power to run it. > > > > This is actively maintained and has been going for a while. > > We use it here (on a fairly small scale) and I know that > > others use it elsewhere for bigger things. It's based on > > a threaded socket server so whenever someone starts to say: > > "I know; I'll roll my own threaded socket server", I'm > > inclined to say: "Don't reinvent the wheel; try Pyro". > > > > PyLinda -http://www-users.cs.york.ac.uk/~aw/pylinda/ When I looked at this, it looked very nice too, but different from pyro. > > > > This implements the tuplespace paradigm. It's great > > fun to use, but as far as I know this implementation > > was a PhD project and lacks the robustness and wide > > use of other things. That said, it works perfectly > > well within its remit and might be a good match for > > what you're trying to do. > > > > No doubt other people can chime in with suggestions > > > > TJG > > Possibly, IPython's new interactive parallel environment is what you > are looking for: http://ipython.scipy.org/moin/Parallel_Computing Unqualified to comment on this - Hendrik -- http://mail.python.org/mailman/listinfo/python-list
