jtag bypass: why isn't SCAN-IN to bypassed interfaces controlled?

[Altera_Forum]

In devkits, like the Arria II GX, there are jumpers for bypassing chips or interfaces (MAX II, HSMA, HSMB, etc) in the JTAG chain. 

 These mux bypass the TDO so that the 'bypassed' interface's TDO doesn't affect the chain being shifted out.   But, the TDI, TCK, and TMS still go to the 'bypassed' interface.   Doesn't this mean that JTAG usage can cause unintended effects to the bypassed interface?   (Obviously, if there's nothing connected to something, like HSMC-A, then it needs to be bypassed & there's no risk. But I'm concerned about whether I need to avoid bypassing connected things, including the MAX II chip.)   Am I missing something, or is there really a risk for accidentally shifting something into a bypassed interface/chip and getting undesirable results?   -Brian

[Altera_Forum]

When operating a JTAG chain, the master is required to shift no-op instructions, usually BYPASS, to all devices he doesn't want to address. As far as I'm aware of, the Quartus software sends a '1' sequence in front of an instruction, so "hidden" devices behind the addressed device will be deactivated. This should be always the case, when the first chain devive (the FPGA) is addressed.  

 But if you are operating a JTAG devive behind the MAX II with MAX II bypassed, there's a risk of unintented actions, I think.