Alexandre, C has a fairly unique place among languages. We could debate how far that goes and the reasons for it, but a few that come to mind are: it arrived a long time ago; it was widely adopted; *nix lives by it. Pharo has to speak it.
In most situations where I have used Dolphin's COM calling capabilities, it was either as a quick way to make something work (type libraries are pretty cool and Dolphin's type library analyzer is even cooler), or part of a desperate project-saving maneuver. As an example of the latter, I have some wrappers that do some very basic things with Word and Excel. They're ugly, but they work in a pinch. Beyond some simple device interfacing (e.g. A/D board), COM has not been of much use. COM you ask - I mean .NET, right? Don't get me started ;) Can we declare the vision of the software IC to be a complete failure? I submit that it did fail, and rather miserably. Further, I will suggest a reason for its failure: interfaces are difficult to get right, and most IDL specs I have encountered are proof of that, being cumbersome, inefficient, and never quite at the right granularity. There are exceptions, but most COMponents appear to be based on the first IDL design that didn't crash followed by a rush to market. Result == rubbish. Despite all of that, and the hideous condition of the MS Office IDL in particular, it did all come together to save my skin. Others might disagree, but I suspect a Java interface would fill a similar role: it would probably be limited to giving a rarely-used but much-needed bridge to something big. OpenOffice comes to mind, and there are probably others. When I need to do something in a static language (usually tight loops for number crunching or graphics), I end up working in C++ with extern "C". The no doubt makes my interest in C FFI all the larger, but I still maintain that it is the default. Anyone who turns to Java will of course find additional use for a Java interface. Like COM, I suspect Java would be something I would try to avoid using, but could find very useful in a pinch. Does that help? I'm trying to encourage you without overestimating the size of the "market." Bill Wilhelm K. Schwab, Ph.D. University of Florida Department of Anesthesiology PO Box 100254 Gainesville, FL 32610-0254 Email: bsch...@anest.ufl.edu Tel: (352) 273-6785 FAX: (352) 392-7029 >>> alexandre.ber...@inria.fr 12/10/08 2:08 PM >>> Dear List, Something is trotting in my head for few weeks. According to the mails exchanged on this list, it seems that interacting with C is of a high priority. I was wondering whether you had a similar need for talking to the Java world. Few months ago, I worked on Athena, a small Smalltalk VM written in Java. It can be interfaced with Squeak. This means that within Squeak, you can create Java objects and talking directly to them within Squeak. Is there anyone who need this? I am ready to continue on this, but I would like to be use case driven. Cheers, Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. _______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project _______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
