view nt/gmake.defs @ 77663:32334cd7302e

(python-guess-indent): Check non-nullness before comparing indent against the 2..8 interval.
author Stefan Monnier <monnier@iro.umontreal.ca>
date Tue, 08 May 2007 01:35:26 +0000
parents 3d45362f1d38
children 713ed3402287 95d0cdf160ea
line wrap: on
line source

#  -*- Makefile -*- definition file for building GNU Emacs on Windows NT.
#  Copyright (C) 2000, 2001, 2002, 2003, 2004,
#    2005, 2006, 2007 Free Software Foundation, Inc.
#
#  GNU Emacs is free software; you can redistribute it and/or modify
#  it under the terms of the GNU General Public License as published by
#  the Free Software Foundation; either version 2, or (at your option)
#  any later version.
#
#  GNU Emacs is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#  GNU General Public License for more details.
#
#  You should have received a copy of the GNU General Public License
#  along with GNU Emacs; see the file COPYING.  If not, write to
#  the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
#  Boston, MA 02110-1301, USA.

# Ensure 'all' is the default target
all:

# NOTES
#
# I tried to force gmake to use the native shell for simplicity, by
# setting SHELL as below, but this didn't work reliably because of
# various case sensitivity niggles.  Specifically, COMSPEC (which is in
# fact usually spelled ComSpec on NT, to make life difficult) typically
# references "cmd.exe" (on NT) when the file is actually called
# "CMD.EXE" on disk for hysterical raisons.  As a result, GNU make
# thinks it doesn't exist (unless compiled with a switch to ignore
# case), and so doesn't change which shell it will invoke to execute
# commands.
#
# It would be possible, though very tedious using just gmake facilities,
# to convert the COMSPEC value to uppercase to solve this problem, but
# it isn't worth it.  That is partly because, even when using the native
# shell, gmake tends to not be happy with backslashes in command
# strings.  The obvious solution is to use forward slashes as much as
# possible, which can be made to work most of the time (putting
# filenames in quotes often helps), but there are still some internal
# cmd.exe commands like `del' and `copy' that won't work with them.
# Although it is possible to convert slashes to backslashes when
# necessary, gmake requires explicitly calling its subst function, which
# nmake does not understand).  For this reason, it is simplest to
# mandate that rm and cp be available, so we can use Unix-format file
# names everywhere.  (Fortunately both MS and GNU make, and the
# respective compilers, are happy with Unix-format names.)
#
# Since we cannot easily force the choice of a particular shell, we must
# make the effort to cope with whichever shell is being used.
# Fortunately, the only command we need to use that is shell specific is
# the testing of a file's existence for the purpose of working out when
# we are copying files to their original location.  That particular
# requirement is abstracted easily enough.
#
# The only other problem area was the change of directory when running
# temacs to dump emacs.exe (where gmake doesn't support cd foo in any
# useful way), but that has been resolved by modifying the Windows
# unexec function slightly to not require the directory change while
# still allowing objects and binaries to be in subdirectories.

# This doesn't work.
#SHELL:=$(COMSPEC)

# Determine whether make is using sh or cmd/command as shell; cmd.exe
# will output "ECHO is on" when echo is given by itself, while sh will
# not produce any output.
sh_output := $(shell echo)
ifeq "$(findstring ECHO, $(sh_output))" "ECHO"
THE_SHELL = $(COMSPEC)$(ComSpec)
SHELLTYPE=CMD
else
USING_SH = 1
THE_SHELL = $(SHELL)
SHELLTYPE=SH
endif

MAKETYPE=gmake

# The following "ifeq" does not appear to DTRT, and therefore breaks
# the build on mingw32. Also the -m option does not exist in many
# (reasonably recent even) versions of Cygwin. These issues need to be
# remedied before putting this cygpath kludge back in.

# Convert CURDIR to native file name, if in Cygwin format
#ifeq "$(shell cygpath $(CURDIR))" "$(CURDIR)"
#CURDIR	        := $(shell cygpath -m $(CURDIR))
#endif

THISDIR		= .

# Cygwin has changed quoting rules somewhat since b20, in a way that
# affects makefiles using sh as the command processor, so we need to
# detect which rules to use.
ifdef USING_SH
sh_output := $(shell echo [Please ignore a syntax error on the next line - it is intentional] 1>&2)
sh_output := $(shell echo foo")
ifeq "$(sh_output)" ""
NEW_CYGWIN = 1
endif
endif

ALL_DEPS	= $^
EMPTY =
SPACE = $(EMPTY) $(EMPTY)

SUBSYSTEM_WINDOWS=-Wl,-subsystem,windows
SUBSYSTEM_CONSOLE=-Wl,-subsystem,console

# INSTALL_DIR is the directory into which emacs will be installed.
#
ifndef INSTALL_DIR
INSTALL_DIR     = $(CURDIR)/..
endif

export EMACSLOADPATH

# Determine the architecture we're running on.
# Define ARCH for our purposes;
# Define CPU for use by ntwin32.mak;
# Define CONFIG_H to the appropriate config.h for the system;
#
ifdef PROCESSOR_ARCHITECTURE
# We're on Windows NT
CPU		= $(PROCESSOR_ARCHITECTURE)
CONFIG_H	= config.nt
OS_TYPE		= windowsnt
 ifeq "$(PROCESSOR_ARCHITECTURE)" "x86"
ARCH		= i386
CPU		= i386
 else
  ifeq "$(PROCESSOR_ARCHITECTURE)" "MIPS"
ARCH		= mips
  else
   ifeq "$(PROCESSOR_ARCHITECTURE)" "ALPHA"
ARCH		= alpha
   else
    ifeq "$(PROCESSOR_ARCHITECTURE)" "PPC"
ARCH		= ppc
    else
error Unknown architecture type "$(PROCESSOR_ARCHITECTURE)"
    endif
   endif
  endif
 endif
else
# We're on Windows 95
ARCH		= i386
CPU		= i386
CONFIG_H	= config.nt
OS_TYPE		= windows95
endif

AR		= ar -rsc
AR_OUT		=
CC		= gcc
CC_OUT		= -o$(SPACE)
LINK		= gcc
LINK_OUT	= -o$(SPACE)
RC		= windres -O coff
RC_OUT		= -o$(SPACE)
RC_INCLUDE	= --include-dir$(SPACE)

libc		=
baselibs	=
O		= o
A		= a

BASE_LIBS	= $(libc) $(baselibs)

ADVAPI32	= -ladvapi32
COMDLG32	= -lcomdlg32
GDI32		= -lgdi32
MPR		= -lmpr
SHELL32		= -lshell32
USER32		= -luser32
WSOCK32		= -lwsock32
WINMM     = -lwinmm
WINSPOOL	= -lwinspool

ifdef NOOPT
DEBUG_CFLAGS	= -DEMACSDEBUG
else
DEBUG_CFLAGS	=
endif
CFLAGS          = -I. -DWIN32_LEAN_AND_MEAN -D_WIN32_WINNT=0x0400 $(ARCH_CFLAGS) -D$(ARCH) \
		  -D_CRTAPI1=_cdecl \
		  $(DEBUG_CFLAGS) $(USER_CFLAGS) $(LOCAL_FLAGS)
EMACS_EXTRA_C_FLAGS = -DUSE_CRT_DLL=1

# see comments in allocate_heap in w32heap.c before changing any of the
# -stack, -heap, or -image-base settings.
TEMACS_EXTRA_LINK = -Wl,-stack,0x00800000 -Wl,-heap,0x00100000 -Wl,-image-base,0x01000000 $(SUBSYSTEM_CONSOLE) -Wl,-entry,__start -Wl,-Map,$(BLD)/temacs.map

ifdef NOOPT
OBJDIR          = oo
else
OBJDIR          = oo-spd
endif
$(OBJDIR):;	-mkdir "$(OBJDIR)"
BLD             = $(OBJDIR)/$(ARCH)
stamp_BLD:      $(OBJDIR)
		-mkdir "$(BLD)"
		echo $(BLD) > $@

COMPILER_TEMP_FILES =

CP		= cp -f
CP_DIR		= cp -rf
DEL		= rm
DEL_TREE	= rm -r

ifdef USING_SH

IFNOTSAMEDIR	= if [ ! -s ../same-dir.tst ] ; then
FOREACH		= for f in
FORVAR		= $${f}
FORDO		= ; do
ENDFOR		= ; done
ENDIF		= ; fi
ARGQUOTE	= '
ifdef NEW_CYGWIN
DQUOTE		= "
else
DQUOTE		= ""
endif

else

IFNOTSAMEDIR	= if not exist ../same-dir.tst
FOREACH		= for %%f in (
FORVAR		= %%f
FORDO		= ) do
ENDFOR		=
ENDIF		=
ARGQUOTE	= "
DQUOTE		= \"

endif

ifdef NODEBUG
DEBUG_FLAG =
DEBUG_LINK =
else
DEBUG_FLAG = -gstabs+ -g3
DEBUG_LINK = -gstabs+ -g3
endif

ifdef NOCYGWIN
NOCYGWIN = -mno-cygwin
endif

ifeq "$(ARCH)" "i386"
ifdef NOOPT
ARCH_CFLAGS     = -D_X86_=1 -c $(DEBUG_FLAG) $(NOCYGWIN)
else
ARCH_CFLAGS     = -D_X86_=1 -c $(DEBUG_FLAG) $(NOCYGWIN) $(MCPU_FLAG) -O2 \
		  # -fbuiltin \
		  # -finline-functions \
		  # -fomit-frame-pointer
endif
ARCH_LDFLAGS	= $(SYS_LDFLAGS)
else
ERROR Unknown architecture type "$(ARCH)".
endif

LINK_FLAGS	= $(ARCH_LDFLAGS) $(DEBUG_LINK) $(NOCYGWIN) $(USER_LDFLAGS)

export XMFLAGS

.DEFAULT:

$(BLD)/%.o: %.c
		$(CC) $(CFLAGS) $(CC_OUT)$@ $<

# arch-tag: 35eb9662-8534-4bcf-b891-0730a09d657f