[ntp:questions] Unexpected executable locations from "sudo make install" on RPi Linux
david-taylor at blueyonder.co.uk.invalid
Sat Nov 16 12:26:57 UTC 2013
On 16/11/2013 11:58, Rob wrote:
> Merely upgrading the kernel from 3.2.27 to 3.6.11 will not cause such
> a change. However, it is always risky to do so because there may be
> dependencies between the kernel and other parts of the system, like
> So what people usually do is use "apt-get upgrade" to perform such an
> update, and of course it may change a lot of other things.
> Installation locations are normally determined by "configure" scripts
> that are part of downloaded packages, and use very complicated mechanisms
> to determine the best installation location for programs. They can
> also be directed using options to the "configure" script. Finally
> a Makefile is written that normally is quite static. I would not
> expect a "make install" to install to a different location when the
> system is changed and "./configure" has not been run again, but
> who knows...
No, I'm going to do the complete upgrade using apt-get, and see what
happens. The difference was between a freshly installed 3.2.27 and
freshly installed 3.6.11+ systems (on two different Raspberry Pi cards).
Having just done the upgrade (almost any settings), I will add my own
kernel and modules which have PPS support, and the changes suggested in
NTP bug 2314 (which was actually an OS configuration issue). It looks
to mostly be working as expected, perhaps will the omission of my snmp
extension for CPU temperature monitoring.
Unfortunately I do need to use ./configure --enable-linuxcaps with NTP
to enable the PPS support in NTP (if I recall correctly), so may be that
will mess things up, and my next compile will have the executables
sprayed over the various directories. At least now I know it may
happen, I can be on the lookout for it!
My notes are here:
and are very much a living work!
Thanks for your help.
More information about the questions