Actually, that's not the part used by Rosewill.
It's
cheaper at Digikey - $2.28/ea up to 10, then $2.06 and falling past 10.
But the construction from the top clearly indicates it's not a Molex part. Note that the STOCK part has a straight-cut outer latch. Molex parts have an ANGLED cut to the outer friction latches. It's a hallmark of Molex mini connectors, in fact, because it increases the number of insertion cycles before they stop catching properly. There's also a much, much tighter feeling on Molex parts.
Oh, and it's also
too damn expensive - $2.28/ea? That's pricey as hell, especially when you're trying to make a sub-$100 keyboard where the switches are near a buck each even at quantity. The most likely ACTUAL part used by Rosewill is a Hirose or JAE Electronic part which costs less than half as much and can be sourced locally to save more on freight.
The problem with this design is when it's not perfectly flat, unless you fill the through hole posts with rock hard solder, load gets transferred to the rear pins which are usually NOT designed to be load bearing. What they probably SHOULD be using is a
Wurth Electronics 651005136521 (how's that for a mouthful?) which has four through-hole posts (and still only costs $1.76/ea!) or possibly Molex 54819-0511 which is basically a shallower design version of the 56579-0519 with slightly wider lugs.
Personally I'd love one of the FTDI true bolt-through designs, but those are full USB UARTs and definitely can't be swapped in.
