I know that this propsal will likely be difficult to digest, because it has huge potential but also would likely involve some heartache. But, I will put it forward nonetheless. It's difficult to fully sum up, so I'll just give the big picture here and see if anyone is interested in more information.
Back in the early 90s I began work on a general purpose C++ framework. This work continued very consistently up through the early 00's. At that point, when the internet bubble burst, I decided to go out out on my own. Since then I have gone on to expand it massively and build a huge automation system on top of it, though the two code bases are completely separate. At this point the whole project comes in at just under 800,000 lines of code. This is proven, in the field type code in a very complex, multi-user, network distributed product. So 800K really high quality lines of code. It is kind of a complete soup to nuts world unto itself, starting with an operating system abstraction layer on the bottom, outside of which no language or OS headers appear, so everything is built on to of it purely in terms of it's own interfaces. So the non-GUI parts at least should be quite portable. I work on WIndows but at one time the back end was running on Linux, purely through recompilation, no conditional code. That platform driver code is still there though out of date now. The GUI stuff is also completely encapsulated, but of course that doesn't mean it's necessarily easily portable. Still, even if it were only the back end, that would include a vast amount of highly integrated functionality that really everything in the Apache world could ultimately be built on, and which would provide a hugely powerful platform for those folks using Apache systems could build on subsequently as well. Just to give a brief listing of some of what the general purpose part provides: 1. Full platform abstraction 2. Loadable resource system with resource definition language and compiler 3. All the usual core stuff, strings, memory buffers, bit sets, collections, streams, files, file system, system information, mutexes, events, semaphores, threads, processes, etc... 4. Text transcoding framework 5. XML parser (I wrote the original Xerces C++, and this is sort of based on that, but fully integrated into a powerful OO framework so much nicer, though not as far along as yours.) 6. Graphics file format framework, currently with just PNG, JPEG, and BMP, but easily expandable to others. 7. All the usual socket oriented stuff and serial port stuff 8. An ORB with XML based IDL language and IDL compiler. It's based on the streaming system, so it's very similar to Java RPC type stuff, i.e. very little code generated. 9. Based on the ORB are standard Name, Logging and Configuration servers 10. Image processing 11. CD ripping and audio file format support 12. Metadata management 13. Basic math libraries and some more advanced stuff built on top of that (e.g. vectors, matrices, etc...) 14. A very nice test framework for doing automated unit and regression tests and generating reports. It runs each test in a separate process, for stability. 15. Support for HTML and SMTP. This is a little basic now but easily beefed up. 16. A simple object oriented data base engine for program configuration and such 17. An ODBC wrapper 18. Regular expression engine. Just supports simple expressions now, but could easily be expanded to include more complex expressions. 19. A ZLib compression implementation 20. A virtual machine based object oriented macro language with compiler, virtual machine and graphical IDE, all embeddable for easy use within apps. 21. A cryptography framework, with current support for Blowfish, MD5, AES, and SHA1, plus unique id generation. Easily expanded to include other algorithms. 22. An application shell engine for managing other programs. 21. You may not want to use it, but it has it's own build tool with project definition language. 22. Color space conversion type stuff. 23. A powerful logging system, which can be easily configured to log to the ORB based log server if you want, or to local binary or text files, shared between multiple apps if desired. 24. A general stats maintenance and access system so apps can maintain various status and they can be gotten to in a standard sort of way, including remotely. On the GUI side 22. Wrappers around some basic window types, frame and menu, and all of the core windowing framework that everything else is built on. 23. Custom implementations of all the other window types, based on the above, so they are fully integrated into the system. 24. Stuff like file open dialogs, palette dialogs, calendar, etc... various common dialog types. 25. Wrapper around the web browser control 26. Extension of the loadable resource system for graphics type resources Anyway, there's a lot of stuff. A good bit more than I could cover above. But that should give you some idea of it. And that doesn't include the automation system that is built on top of it. Keep in mind that, with just a few small exceptions, everything is a custom implementation, built in terms of this framework, so that they will be portable. The exceptions are the JPEG stuff, which uses a bit of the publically available JPEG library, the WMA codec stuff, temporarily the MP3 metadata stuff which currently uses a Windows SDK but that could be replaced. The reason I'm doing this is because, for various reasons having nothing to do with technical excellence, my automation product is not going to be successful in the professional install world it needs to in order to be a viable company. So, I'm going to be be in need of employment and I promised I would open source the product if I even had to give it up. So this might be a way of doing both. If Apache is interested in the extremely powerful automation product, that would sort of come for free since it's built on top of the above framework. It is as large as the general purpose code base. You can see it at www.charmedquark.com. So, there it is. I know that's a lot to digest. And, even if you accepted it, there would be a LOT of politics in terms of really getting the best use of it, which really involve using it as the basis for other Apache products. The level of integration it could provide would be massive. It's someting that really hasn't existed for the C++ world. I know that there have been some libraries out there. But I don't think really anything like this, where it was built by a single person with almost no commercial pressures for most of its life. So it has a level of quality and consistency that you wouldn't like find elsewhere. And also because almost none of it is just a wrapper around some black box of code. So almost everything single thing is debuggable and should be portable (other than noted above.) If there's any interest let me know. ------------------------ Dean Roddey Chairman/CTO Charmed Quark Systems, Ltd <http://www.charmedquark.com> www.charmedquark.com
