For everyone running the devel version, I think I'm almost through with coding ascent issues.
* the helium system should now be mostly realistic, there's a detailed leakage simulation distinguishing isolatable from non-isolatable leaks and the interaction between different tanks should also be okay according to the valve schematics The only thing that's not realistic is that manifolds bleed down 'immediately' when isolation valves are closed - this doesn't really affect procedures, but I am aware that they don't do that in reality
* I've added a solid bunch of CWS messages - in particular command path and data path failures as well as more button failures (ETSep and Cutoff buttons may now fail) - as well as the failures of course. Mostly the failures are also realistic - a data path failure will 'hide' other issues which may or may not exist, so to correctly deal with this you need to understand the MPS and procedures (or you cheat and check whether then engine is still running in outside view...)
Implementing command and data path failures required me to re-do the way engine throttling and cutoff is handled - I *think* I've done it correctly for all scenarios (aborts included), but if you discover a scenario where no fault is present and yet an engine is not throttled, it's probably my fault
* in addition, an engine can now run 'redline' - which means the engine controller deems it unsafe to run. The limit enable/inhibit switch now governs what the controller will do - if limits are enabled it will shut down the engine, i.e. this is equivalent to a failure, but if limits are inhibited it will continue to run the engine. This gives a situation with a certain failure probability per time -
most of the time the affected engine will just break when it finally dies, but
sometimes it will blow the ET in the process. Nevertheless, there are situations where it is preferable to keep a redlined engine running - flying under BFS comes to mind, since the software can't handle a single-engine situation.
* there's also a couple of seemingly weird indicator couplings implemented - for instance if you get a GPC sync failure during ascent, it may not only vote the GPC out of the set but also create the indication of an engine lockup - which however is not real.
I've introduced the new concept of 'training scenarios' into the menu - (easy and medium already work) - basically by selecting one scenario while on the pad, you'll get a random failure of something during ascent. Easy fails just a single system - so you might lose an engine or deal with a helium leak - medium is more involved and gives you a staggered failure with more than one component breaking - hard will get the tough cases where you really need to analyze a situation correctly (and quickly) in order to survive.
So that's what you get for X-mas to play with