I have been muttering about SBCL runtime cleanups lately#

Thu Mar 24 21:52:48 2005

I have been muttering about SBCL runtime cleanups lately. Current thoughts are

• smaller source files which do one (or a few) things each: e.g vanilla-cheneygc-sigsegv-handler.c

• dispense with Config.`uname m`, instead choosing the files to compile/link based on *targetfeatures - either in the Makefile directly if we can find a sensible way to export target-features* to the shell, or with a big features.h file that consists entirely of

  if defined(LISP_FEATURE_SB_THREAD) && !defined (LISP_FEATURE_SB_PTHREAD)
  include "sigprocmask-pthread.c"
  endif
  [...]
  

  etc. As a general principle, though, #including *.c files gives me the creeps, though, so it's more liekly we hack genesis to write out something that make(1) can grok.
• probably also implement the long-threatened separation of build features from target features. My current idea for this is that when started from a warm core we just create features from (what was
  • target-features* at build time) then delete from it wherever we find we built with support for a feature that's not present at runtime. I think that Christophe has an arguably neater hack involving the package system, though.

Depending on work workload, and on how long it takes to get acpi suspend working properly on my new laptop (yes, I finally gave in - it's a compaq nc4000; not quite as small as its predecessor but still reasonably light), hope to see something along these lines early next month.