On Wed, 28 Nov 2001 13:16:08 -0500, "Eric S. Raymond" <[EMAIL PROTECTED]> wrote: >Keith Owens <[EMAIL PROTECTED]>: >> # make oldconfig >> # grep BLUEZ .config config.out >> .config:# CONFIG_BLUEZ is not set >> config.out:CONFIG_BLUEZ=n >> # make oldconfig >> ... >> BLUEZ: Bluetooth subsystem support < >: (FROZEN) >> BLUEZ_HCIUSB: HCI USB driver < > (NEW)?: >> >> Why does the second make oldconfig get rid of BLUEZ_L2CAP and friends? > >Because `make oldconfig' reads in BLUEZ and freezes it to n. >Symbols with ancestors frozen to n aren't written to the config file. >This is how CML2 avoids writing symbols associated with (say) SPARC when >you're configuring for X86 -- SPARC gets frozen to n. > >The real design question here is whether make oldconfig should freeze symbols >read in or merely set them without freezing them.
IMHO make oldconfig should always output _exactly_ the same .config as was read in, apart from changes due to new options. CML1 does that and some people rely on this behaviour to detect new options, by comparing the old and new .config. I was surprised that the first pass generated descendents of BLUEZ when BLUEZ==n. _______________________________________________ kbuild-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/kbuild-devel
