Subject: Planning MonoDevelop 1.0
Date: Wednesday 22nd November 2006 23:28:39 UTC (over 10 years ago)
Hi all! After several years of development MonoDevelop is beginning to be really useful to develop real applications, so I think it makes sense to start planning a stable 1.0 release. My proposal is the following: * MD beta 1 release around mid January (two months from now). In this release we would include most of the missing features for 1.0 (see below). We would do code freeze at this point. * MD beta 2 (or RC) release around mid February. Mainly bug fixes and memory usage reduction. * MD 1.0 release around mid March. It will include the latest bug fixes. I also think we should work on writing documentation for this release, specially tutorials. So, which features is MD 1.0 going to include?. I'd like to get input from contributors and users about that, but I think what we should do is to finish and polish what we already have. I went through all add-ins and I have written down some notes about what's missing on each of them (bugs aside): * MonoDevelop core: The core can be considered complete now. No major new features are planned. However, we need to make an audit of the API to make sure we don't have public classes or methods which don't really need to be public. I'd like also to find some time to reorganize a bit the extension point tree (which has a mixture of SharpDevelop and MonoDevelop branches), to something like what I described some months ago: http://lists.ximian.com/archives/public/monodevelop-list/2006-January/003077.html * Debugger: I know a lot of people will ask for this, but I fear the debugger won't be ready by our code freeze date. We can release it later on as a separate add-in. * Documentation browser: The Monodoc integration more or less works, but there are some known crashes. However I still believe that showing Monodoc in an external window would be more practical that the current integrated view. The integrated documentation tree is not really useful, we don't have search or an index integrated. * Text editor: There are several requests/issues to solve: * Parameter completion: after typing the name of a method, show a popup window with information about the parameters the method takes. * Remove C#-isms. The code completion engine has several C#-isms in the code, which may make it difficult to integrate with other languages. * Configurable key bindings. That's not specific to the editor, but people mostly complain about it. I'm not sure we can implement this for 1.0. * Refactory operations. We have and under-used refactory API. It would be nice to use it to implement some basic refactoring operations. * VS importer: Ankit is working on supporting VS2005 files right now. The idea is that you can open VS files directly in MD and work with those projects like if they were native projects. It will support writing changes such as adding files or modifying the project configuration. Ideally, we should do the same for VS2003 (right now it only supports importing for VS2003 projects). * ASP.NET support: I don't know if this add-in can be mature enough for the 1.0 release. I'll need input on that (which features could we release? would the add-in be really useful for real world web projects?). * GTK# designer: There are several features to be implemented in the designer before the release: * Support partial classes when generating code. * Support for Undo/Redo. * Reuse the toolbox implemented for the ASP.NET add-in. * Support for some missing controls, such as Gtk.Fixed and context menus. It would be nice to have a TreeView designer, but it may be too much work for 1.0. * MonoQuery database management: I have no idea of what is the status of this add-in. I have never really used it. Can somebody give input about it? Do we want to support it for 1.0? * NUnit: It can be considered feature complete for 1.0. The only missing issue to resolve is what to do with the NUnit dependency. Right now the add-in can't be built with recent NUnit versions since they have breaking changes. I talked about this with Charlie Poole in the Mono summit and he suggested to embed the most recent NUnit dlls into the add-in, since even if the API has some breaking changes, new NUnit versions can still run tests written for all the previous versions. * Version Control: I recently integrated some of the changes done for the Google SoC project and improved it a lot. Support for SVN is mostly functional. What's missing is support for CVS. It would be nice to have for MD 1.0. * Welcome page: It looks ok to me. * Web references: I'd like to integrate the web references add-in implemented by Gideon de Swardt. The only problem I see is that it only supports .NET 2.0 projects. It would be nice to make it work for 1.1 as well. * Xml editor: It's being maintained outside of MD, so nothing to add. Just to make sure it works when we release 1.0. * Boo: I think this add-in is complete, but I'd also like to know the opinion of Peter about this. In any case, there is one issue that should be fixed: right now the Boo add-in always starts an external process which runs the backend for the Boo shell. This process should only be started if the user opens the shell, since it's a waste of memory otherwise. * CSharp: It is mostly complete. We may need to add support for partial classes. I'm not sure they are correctly handled by the parser right now. * ILAsm: I haven't used it for a long time. I guess it works. * Java: It's mostly complete, although there are some IKVM configuration issues that need to be fixed. * VB.NET: I don't think there is support for generics. We may want to migrate to the new VB compiler. * Nemerle: Don't know much about the status. Any input? The big problem I find is that it depends on a specific version of Nemerle from SVN. If it can't be made to depend on a packaged Nemerle version we won't be able to include it in the MD release. * ChangeLog add-in: It's now integrated with the version control add-in. There are ideas for improving it. * Autotools: There are several features I'd like to support for MD 1.0: * Support integration with existing Makefiles. That is, if somebody has a project with complex build rules which can't be emulated by MD, just make it possible to delegate the entire build process to Make, and still be able to launch the build from inside MD. About modifying the Makefile in MD, I think supporting adding/removing files would be enough for now. * We need to find a solution for file templates (.in files). MD uses some of them. We need to support them in the MD build model if we want to be able to completely build MD using MD. * There may be other missing features needed for self-hosting MD. * It would be nice to support generating a tarball without having to pollute the project directory with autotools files. * IDE extensions for Mono: Nothing much to do. Just make sure it works with the most recent Mono makefiles. As you can see that's a lot of work. I think we should prioritize working on features related with development GNOME desktop applications (which has been the main goal of MD anyway). I'd like to get feedback about this plan. I may not have a complete view of every add-in, and I'm sure I'm missing things. Thanks! Lluis.