[Python-Dev] _sre.c's compiled size

Thomas Wouters thomas@xs4all.net
Fri, 7 Jul 2000 13:56:41 +0200


--5vNYLRcllDrimb99
Content-Type: text/plain; charset=us-ascii


For the record ;) I've compiled _sre with -fno-inline, on a RH6.2 egcs-1.1.2
system, and I have to admit compilation is going a lot faster. About 1/10th
of the time, compared to without -fno-inline, which means it's about 1/100th
of the annoyance when compiling a lot ;-) Also, the resulting object file is
only 71K, instead of 440K, on my machine.

Overall speed of python, as measured by pybench, doesn't really change, but
pybench does not measure regexp speed at all, and I don't have a suitable
testpackage for that.

I've attached a teensy patch that allows '-f' flags to be used in
Setup files, to be able to easily switch between the two.

-- 
Thomas Wouters <thomas@xs4all.net>

Hi! I'm a .signature virus! copy me into your .signature file to help me spread!

--5vNYLRcllDrimb99
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="makesetup.patch"

*** src/Modules/makesetup	Thu Jul  6 13:58:06 2000
--- src-rangemaker/Modules/makesetup	Fri Jul  7 13:23:25 2000
***************
*** 140,146 ****
  			srcs)	srcs="$srcs $arg"; skip=; continue;;
  			esac
  			case $arg in
! 			-[IDUC]*)	cpps="$cpps $arg";;
  			-Xlinker)	libs="$libs $arg"; skip=libs;;
  			-rpath)		libs="$libs $arg"; skip=libs;;
  			-[A-Zl]*)	libs="$libs $arg";;
--- 140,146 ----
  			srcs)	srcs="$srcs $arg"; skip=; continue;;
  			esac
  			case $arg in
! 			-[IDUCf]*)	cpps="$cpps $arg";;
  			-Xlinker)	libs="$libs $arg"; skip=libs;;
  			-rpath)		libs="$libs $arg"; skip=libs;;
  			-[A-Zl]*)	libs="$libs $arg";;

--5vNYLRcllDrimb99--