[Bf-committers] NAN_DEPEND in makefiles...

Hans Lambermont hans at lambermont.dyndns.org
Mon Apr 18 20:22:11 CEST 2005


Ton Roosendaal wrote:

> (Well, I hope Hans is listening too!)

/me wakes up

> Quite some of Windows testing builds get created with cygwin, using  
> makefiles, and I've noticed that these builds tend to be instable,  
> unless compiled over after doing a full 'make clean'.

sounds like ...

> The Windows section of nan_compile.mk contains a commented out  
> "NAN_DEPEND = true".

this ;-)

> This smells to me like the main cause of problems.  
> So, is this true, and if so why is it disabled?

Good question. cvs annotate shows this :
    1.6          (hos      25-May-03):     #NAN_DEPEND = true
with cvs log for 1.6 :
    revision 1.6
    date: 2003/05/25 05:22:46;  author: hos;  state: Exp;  lines: +41 -45
    Some tweaks to make the windows-gcc play nice with irix.
That's weird. This is inside this :
    ifeq ($(OS),windows)
	ifeq ($(FREE_WINDOWS),true)
Doesn't sound like something 'irix' ;-)

cvs diff -r 1.5 -r 1.6 nan_compile.mk shows :

    Index: nan_compile.mk
    ===================================================================
    RCS file: /cvsroot/bf-blender/blender/source/nan_compile.mk,v
    retrieving revision 1.5
    retrieving revision 1.6
    diff -u -w -r1.5 -r1.6
    --- nan_compile.mk      24 May 2003 20:04:20 -0000      1.5
    +++ nan_compile.mk      25 May 2003 05:22:46 -0000      1.6
    @@ -1,5 +1,5 @@
     #
    -# $Id: nan_compile.mk,v 1.5 2003/05/24 20:04:20 hos Exp $
    +# $Id: nan_compile.mk,v 1.6 2003/05/25 05:22:46 hos Exp $
     #
     # ***** BEGIN GPL/BL DUAL LICENSE BLOCK *****
     #
    @@ -180,7 +180,6 @@
                    CFLAGS += -pipe -mno-cygwin -mwindows
                    CCFLAGS += -pipe -mno-cygwin -mwindows
                    CPPFLAGS += -DFREE_WINDOWS
    -               #CPPFLAGS += -I/usr/include/mingw
                    REL_CFLAGS += -O2
                    REL_CCFLAGS += -O2
                    #NAN_DEPEND = true
    @@ -263,9 +262,6 @@
         else
            $(CCC) -c $(CCFLAGS) $(CPPFLAGS) $< -o $@
         endif
    -
    -#$(DIR)/$(DEBUG_DIR)%.res: %.rc
    -#      $(SRCHOME)/tools/cygwin/cl_wrapper.pl - rc /fo$@ $<

     $(DIR)/$(DEBUG_DIR)%.res: %.rc
     ifeq ($(FREE_WINDOWS),true)

which is also not Irix-related, so I guess this is an accidental commit.
Hos was probably committing something else -that was irix related- and
this got pulled in. Now only if we had Bonsai set up, then we could
easily see what other files he committed with this.

However, the diff part above doesn't have the NAN_DEPEND line changed,
so only spacings were changed here (we do diff -w) Therefore some
previous commit was the real commentor of NAN_DEPEND. It appears to be
1.5, also by Hos :

    revision 1.5
    date: 2003/05/24 20:04:20;  author: hos;  state: Exp;  lines: +49 -23
    Support for building bf-blender under windows with gcc (huge thanks to
    Florian Eggenburger).

    Full instructions are in doc/README.windows-gcc.

    Main differences from Florian's patch:

    - the 'lib' dir should now be the same level as the 'blender' dir
      (rather than being a subdir of 'blender'). This is consistent with
      the other platforms that bf-blender supports (tuhopuu will also adopt
      this convention hopefully soon).

    - the script 'free_windows-env.mk' is no longer needed ... see the
    docs about how this is overcome (again, tuhopuu will hopefully
    also follow this route soon).

    - the dlltool dir has it's own Makefile that builds all of the
    needed stub libraries from the dll's in cvs.

cvs diff -r 1.4 -r 1.5 nan_compile.mk
    Index: nan_compile.mk
    ===================================================================
    RCS file: /cvsroot/bf-blender/blender/source/nan_compile.mk,v
    retrieving revision 1.4
    retrieving revision 1.5
    diff -u -w -r1.4 -r1.5
    --- nan_compile.mk      19 Dec 2002 22:06:32 -0000      1.4
    +++ nan_compile.mk      24 May 2003 20:04:20 -0000      1.5
    @@ -1,5 +1,5 @@
     #
    -# $Id: nan_compile.mk,v 1.4 2002/12/19 22:06:32 maarten Exp $
    +# $Id: nan_compile.mk,v 1.5 2003/05/24 20:04:20 hos Exp $
     #
     # ***** BEGIN GPL/BL DUAL LICENSE BLOCK *****
     #
    @@ -174,6 +174,24 @@
     endif

     ifeq ($(OS),windows)
    +       ifeq ($(FREE_WINDOWS),true)
    +               CC  = gcc
    +               CCC = g++
    +               CFLAGS += -pipe -mno-cygwin -mwindows
    +               CCFLAGS += -pipe -mno-cygwin -mwindows
    +               CPPFLAGS += -DFREE_WINDOWS
    +               #CPPFLAGS += -I/usr/include/mingw
    +               REL_CFLAGS += -O2
    +               REL_CCFLAGS += -O2
    +               #NAN_DEPEND = true
    +               #OPENGL_HEADERS = /usr/include/w32api
    +               OPENGL_HEADERS = ./
    +               AR = ar
    +               ARFLAGS = ruv
    +               ARFLAGSQUIET = ru
    +               WINRC = $(wildcard *.rc)
    +               RANLIB = ranlib
    +       else
         CC = $(SRCHOME)/tools/cygwin/cl_wrapper.pl
         CCC = $(SRCHOME)/tools/cygwin/cl_wrapper.pl
         JAVAC = $(SRCHOME)/tools/cygwin/java_wrapper.pl -c
    @@ -197,6 +215,7 @@
         ARFLAGSQUIET = ru
         WINRC = $(wildcard *.rc)
     endif
    +endif

     ifeq (debug, $(findstring debug, $(MAKECMDGOALS)))
         export DEBUG_DIR=debug/
    @@ -245,8 +264,15 @@
            $(CCC) -c $(CCFLAGS) $(CPPFLAGS) $< -o $@
         endif

    +#$(DIR)/$(DEBUG_DIR)%.res: %.rc
    +#      $(SRCHOME)/tools/cygwin/cl_wrapper.pl - rc /fo$@ $<
    +
     $(DIR)/$(DEBUG_DIR)%.res: %.rc
    +ifeq ($(FREE_WINDOWS),true)
    +       windres $< -O coff -o $@
    +else
            $(SRCHOME)/tools/cygwin/cl_wrapper.pl - rc /fo$@ $<
    +endif

     $(DIR)/$(DEBUG_DIR)%.class: %.java
         ifdef JARS

This is FREE_WINDOWS related, but it doesn't explain why NAN_DEPEND was
commented out. I hope Hos still knows why ;-)

Hans
-- 
http://hans.dse.nl/   () ASCII-ribbon campaign against vCards,
                      /\ HTML-mail and proprietary formats.


More information about the Bf-committers mailing list