I compiled the latest RC on my Linux system, here are my findings and issues I noticed:
During compilation, multiple warnings were issued by the compiler (gcc 4.8.1), see this paste: http://pastebin.com/wQwg9BuV
Some of those are harmless (unused variables), some are fine as long as one knows what he is doing (comparing signed and unsigned types), some might be a cause of serious problems (possible use of uninitialized variables, reaching over array bounds, ...)
In-sim issues:
AI/MP LOD ranges - AI aircraft seem to be visible on very short distances only, setting smaller value causes longer visibility range! Default value causes too short visibility. Something is obviously wrong in the range calculation or comparison.
Submodels are visible over insanely short range or not at all. Weird dependance on AI/MP LOD range observed (similar as for AI aircraft), default value causes too short visibility. In case where the submodel is a bullet that is set to explode upon impact and the bullet is out of the visible range, the explosion is visible ONLY with Rembrandt on (there should be no difference between Rembrandt and standard rendering shemes in this regard).
In older versions, submodel visibility range was not limited (AFAIK), only by dist-scale and range animations on the submodel object itself. I think it should remain this way (or have a separate setting). Submodels aren't likely to cause problems because of their limited lifetime, contrary to the AI aircraft, where the range limitiation is obviously needed.
Earthview orbital rendering - When turned on, it confuses the AGL value as seen on the HUD. Also flyby view in dive causes view from below the (fake) surface which is placed very close. At about 100k ft the AGL value was a few hundred ft or less. When the Earthview orbital rendering was turned on on rather low altitudes (I know it in not to be used that way), it could cause aircraft crash into the fake surface.