From: Radu B. R. <ru...@cs...> - 2008-04-25 11:29:16
|
Hi John, John Oliver wrote: > We have been getting a number of these issues on the mailing lists due > to the incompatibility between 2.1 and javaclient2. I read from Leo > Nomdedeu that a solution is being worked on. Could I ask what the form > of this solution is? Due to the fact that player seems to change its > protocols quite often if we change the classes to fit the current spec > is it not quite likely we will be back in the same position in a year > or so when things change again. The current incompatibilities are due to the fact that Javaclient2 was written for Player 2.0.x. The current changes in Player 2.1 were long awaited by anyone who ever wanted to design a custom interface, or, simply reduce the amount of bandwidth/data it had to transfer (e.g. camera images where always an issue with Player - not anymore! :) ). Leo is bravely taking a hit at upgrading Javaclient2, and I will try to help maintain it more in the near future as well. It would be great if more people could contribute though - especially since we seem to have quite a few users out there. We don't need to worry about more incompatibilities and changes between Player and Javaclient. We're not going to change the current Player interfaces anytime soon, so once we upgrade to 2.1 we should be fine for some (long) time. > I notice that the pack functions for the communications in player are > generated at compile time using a python script (playerxdrgen.py). > Could we not modify this to allow us to generate the java side too thus > allowing us to keep player and Javaclient in sync without manually > making the changes? That's what Leo is working on. So far we had to modify and test each interface manually, and that worked fine at the beginning since their number was "maintainable". Nowadays we just have too many, and a scriptable solution which generates them automatically is the right choice. > Unfortunately I do not have too much time to attempt this myself, > however I was wanting to gauge reaction to see if this was the current > approach or if anyone thinks this is possible. > > Thanks > John Oliver I know how that feels. I'm trapped in the same loophole... Still, e-mailing ideas and suggestions does help. Cheers, Radu. -- | Radu Bogdan Rusu | http://rbrusu.com/ | http://www9.cs.tum.edu/people/rusu/ | Intelligent Autonomous Systems | Technische Universitaet Muenchen |