Looks like the "gcc version used" string requires a decimal: "4.0.1" works; "4" does not. Unfortunately, the default setting of "?" also fails this test so if you try to update a bug already in the system, then it won't pass validation and prevents you from saving your change...
The chewing sound you hear is that of an assumption biting me in the ass
I think of gcc versions in 3 digit terms and set 3 characters as the minimum. Which would only enforce 3.*, 4.* etc anyways... it's set to one character minumum now.
I'm still seeing defaulting of processor precision to 64-bit too.
That was me not formatting the enumerations correctly. fixed.