> I'm undetermined. ;-) > Well, since implementation is always the smallest part of the work -- yes, > BMimeType is quite big and the registrar also has to be done, but the > implementations shouldn't be that complex -- we won't get around ending up > both writing the tests. If one of us writes the registrar -- the complete > one or only a skeleton -- doesn't really matter, it will be done much > faster than the tests. We could as well do it the other way around and > finish the documentation and the tests and start with the registrar and > the implementation thereafter. I think I'd prefer the latter choice (tests first, implementation second). Better to get a good feel for how things are supposed to work. > However we start, we should divide the work by functionality first. So > what functionalities do we have: > * basic MIME string stuff (constructor/destructor, Contains(), > SuperType()...) > * database access (Install(), Delete(), GetAppHint(), GetIcon(),...) > * monitoring > * C functions (create_app_meta_mime(), get_device_icon(), > update_mime_info()) And I would also include: * Sniffer functionality (GetSnifferRule(), SetSnifferRule(), CheckSnifferRule(), GuessMimeType()) > I think, I would do the monitoring part, since I can reuse some things > from BQuery for that. The C functions (save get_device_icon()) may be > hard to test. I would take them nevertheless. And the MIME string > functionality. Then the main part remains for you. ;-) > We can bargain though. :-) That sounds good to me. :-) I'd be willing to take the sniffer functionality too, or to leave it to whoever finishes the rest of their tasks first. Either way is fine with me. -Tyler