well, i think this is a matter of understanding what the hell "CVS" actually is, and since i never saw anyone explain it, here's my "normal english" breakdown of it.
the process of "making a program" is fairly complicated and convoluted, but to break it down in a basic way, you get a development environment, and inside of there you write, compile, and test/debug source code. source code in and of itself is NOT a program, at least not in the sense people are used to thinking about. the actual files that you can use (the .XBE file) are compiled from the source code by a compiler, which is basically a program which reads the source code and builds the actual program file out of that source code. to think of it another way, the compiler takes the code that humans can understand and turns it into code that the CPU of the XBox can understand.
that all isn't actually directly involved in CVS, but is neccessary to understand what CVS is and how its used. in modern programming, things are broken down into small pieces to make the program easier to write and understand, as well as allow you to re-use code as much as possible. usually the actual "source code" of a project is broken up into many little files.
now, bearing the above in mind, lets say that you and your friend Jim want to make a program together. CVS is one way of making this easier to do. using a CVS system, the source code for the project is all stored on a server. people who want to can log onto that server and download all the current source. the CVS system keeps track of changes that are made and helps keep the overall project organized, but even so, it's still a bit chaotic. say you have 5 guys all working on the same project. they access the project over the internet and thus don't neccessarily work together following any schedule or anything, so you cannot reasonably expect changes to be made in a regular orderly fashion. instead, say Jim decides he's gonna fix a bug in some part of the project. once he's done he posts it to the CVS and the bug fix becomes part of the project, with a note added into the changelog for the project.
as a result of the above, you can probably understand that an massive and complicated project which is under development like XBMC has changes occuring fairly frequently.
the builds that MOST people use are the T3CH builds, which are simply built off whatever the current CVS is. like i said earlier, when you have a bunch of people working independently, you occasionally have bugs creep in or features that people did not yet have time to 100% finish yet, and the inconsistency of the XBMC CVS builds is a result.
all this really means for you is you're gonna either have to live with the current CVS builds or use only the last "stable" build, which is from a long ass time ago and is missing a ton of new stuff. i hope all that made sense and was illuminating for some people (IMG:
style_emoticons/default/smile.gif)