Slashdot

I tried using XBMC on an Asrock ION 330 as a frontend for a while. Basically, it looks amazing (especially compared to Myth 0.21), and has some nice things like animations. It was dirt simple to get working with the hardware, including an MCE remote (as in, I basically had to do nothing).

The bad: it's not a DVR at all. It has half-baked myth backend support - in that it is supposed to understand the streams and be able to play content. However, you have to go into a menu item called "Scripts" and then start "Mythtv" from a list there, before navigating to recordings. It has no support for scheduling or doing anything besides playing back recordings. I ended up just making it look directly at the Recordings directory on my myth box and playing back files from there (note, I use a script there to symlink the mythtv recording files to their actual names).

The ugly: Due to the high potential, I started digging in more to see if there was anything I could do to help out, such as work on the myth backend support. What I found is that entire project has been mothballed, and they are working on a grandios rewrite of a generic PVR layer, and then later on top of that will have Mythtv support. Not a TERRIBLE plan, but 1) it's a huge plan, that will take a long time before it is even remotely usable, 2) it means the PVR has to be lowest common denominator support, combined based on what all the PVR backends they support have. It also means the devs are rejecting patches to the existing myth support, because it is not relevant in the wake of the new PVR backend.

On top of that, the architecture is sadly lacking. With apologize to XBMC devs, as I'm about to call your baby ugly, but It very much shows its organic and basic roots. The actual menu items are hardcoded into the theme, and intertwined with the code in the back. To do something that should be simple, like add another menu item to the main menu, from what I can tell you have to: 1) modify the core code to understand the command, 2) modify the theme to add in the button - which includes changing the x,y coordinates of all buttons below that one that now need to be shifted, and adjusting the animation code so it knows the positions of all the buttons. It's possible it is simpler than that (I didn't actually try), but from looking at the code, that's what it looked like to me, and so I lost interest due to the amount of effort and non-reusability (eg, my Mythtv button wouldn't be accepted as a patch, and I'd have to redo this anytime I installed an update).

I ran MythTV for six years. The last year I've used SageTV. I got sick of MythTV locking up, crashing, and the constant non-stop twiddling with my configuration because I could never get things quite right.

SageTV isn't much better. I spend a lot less time twiddling, but it crashes and freezes about as often as MythTV used to. I'm still looking for that HTPC that just works. I haven't found it yet.

You sure these aren't hardware-related problems? I've had a dual-tuner, split FE/BE Myth system running for, oh... two years now?... with absolutely no problems. Any crashes I've had occurred early on, and have been hardware related (ie, hard disks failing), or problems with Linux itself (XFS+LVM causing hardlocks, bugs in ivtv resulting in tuners dying, etc). 'course, it helps that once I had a working configuration, I didn't touch it at all (ie, no OS updates, etc).

As for fiddling... honestly, I have no idea what you're doing with your system that requires that kind of care and attention. Again, I've been running a Myth system for two years, and it's required basically zero care and feeding once I got the system up and running and working the way I wanted (granted, that took a bit of time early on, particularly on the frontend, getting third-party software working right, tweaking the remote configuration, etc).