Panel Discussion 2010-02-07 Future of OPenModelica ---------------------- Jan Brugård: Software based on OpenModelica. Main point: core language features, compiler to work well. Main focus on kernel and development tools for the kernel (debugger, etc.) Important the improve the workprocess for the developers. More important to focus on the core first. Olof Johansson. Background in Engineering databases. Siemens, process models, electrical models, integrates with control system design, integration with 3D modeling of power plant. Future: Simulation has been in its own area for a long time. Has been a big gap in have ing a model and transforming it to Modelica. Simulation community, inspiration, come into the engineering tools. The engineering tools will make it easier to generate models. A lot of details that needs to be done before you can do detailed simulation models for a plant. Mission for OpenModelica, spread info about the language, get more engineerings involvved, try to reduce the gap between engineering models and Modelica models. Peter Fritzson and I had some visions 2003, some of these are gradually coming true. Having a platform like OpenMod Francesco Casella Short comments: Process: How OpenModelica is developed. Colleague is starting to be involved in the Linux devepment. - Important the OpenModelica community grows. This has already started, several developers, still very small community. - To have OpenModelica really grow, you need more people Strategic goal. Product: - Supporting non-simulation experts in trouble shooting, solving problems, some parameter is wrong, what should be done to help. - Require research. - Most of the papers you see is about success stories. You don't see papers about how hard it was to get there. - if the goal Oliver Lenord, Bosch-Rexroth. Two sides: - A tool vendor side. We develop software for internal use at BR. The kernel should be stabilized. Happy about the work to modularize and stabilize the kernel. Have too few experts right now. Happy about those experts who we have. COdegen, tearing, jacobians, etc. Everyone do not need to do everything. - User view. Contact with people who try to use this for daylli problem When looking at our company, I see simulation experts, but a much wider range of people who would like to use simulation for applications, but do not have time to set up complex models and do everything from scratch. - When looking on tools that are available, they look almost the same. - Needed to have something that is more applied to a specific problem area, so that you can start right away with applications for an engineer. - Oliver: The design, the engineering is building a bridge. General issue Not directly related to OpenModelica, more general. - PeterF: Per Sahlin has Building simulation domain high level applications. - Jan Brugard: The OpenMdoelica kernel should be stable lenough so that other people can build on top of it for different applications. - Mohsen: Important to modularize the compiler, but also important to have an open source GUI for basic functin Gerhard: 1999 used Modelica; before that Smile. Modeling airplanes, cooling machines - Need reliable tool for the engineering developing process. - Very important to have good debugging tools. Want to model with an open modeling language. - Benchmark. If our models are running, you can buy another tool. - I am not - It should be easy to handle the tool. My focus is the physical engineering process, for this we need an open tool. Kilian Link. 1. I am a user. Developing inhouse tools. Must be possible to do something with the whole toolchain. - Agree with Mohsen that we need a GUI to get started doing something. It is not important that it is as fast as Dymola, but it needs to be useable. I somewhat disagree with others. It is also important to have an open source tool chain. - Jan brugard: Is it important that the toolschain is delivered by the OSMC? - Kilian: It is hopeless for the users. - Jan: Is the OSMC focus to reach the end users or to reach the developers? We cannot do both? - Kilian: It is hard to get companies to put money into it if there is not tool chain or GUI. It should not be hard to get started and - TOmmi Karhela: Software business is moving from licensces to service business We need free toolchain and services. - Kilian: Most important, to get some feedback when something goes wrong. More important than high performance. Debugging and error message is very important. - Kilian: Building a model is always a process of getting to know how the system works. I doubt that there will be database system that - Kilian: How people works. Start with a tiny model. How it works. Then build a bigger model, etc. If you think that you can combine 1000 models - Francesco: Problem if you combine 2-3 models and it still does not work. - Oliver: I agree, for education there whould be a free tool available, so you can educate students. I do not agree that we should invest to have a new Dymola. Therefore, looking at a big company, when convinced your boss that simulation is valuable, you get the money to buy a license. Industry should have the resources. Goal to strengthen the education part, and also provide a solid basis on which others can build on. Solid and robust kernel. - FrancescO. Useful to have credible technology demonstrators. Something that should work for demonstrators. - Jan: - Kilian: If I use an open source tool, why buy a MathModelica license? - Jan: Try to do tools for every single end user is not feasible. Bernhard Bachmann, FH Bielefeld. - I am a Mathematcisian - My interest is to have a chief Dymola simulation, a stable and general backend. - We will focus on further improving the backend, initialization, integration, performance. - Since Modelica is growing I think there will increased interest. - Thanks for Peter for the effort of starting the OpenModelica effort Peter F, Summary: - Stable OpenModelica kernel is needed - Basic graphic user interface is needed for the open source tool to be useable and consortium