[PATCH] kbuild: Make filelist for clean and mrproper distributed 0/6

[PATCH] kbuild: Make filelist for clean and mrproper distributed 0/6

[Sam Ravnborg]

The next six mails contains patches to introduce a distributed way
to specify what files to delete during make clean and make mrproper
The end result and the driver for this patch was the possibility
to get rid of the centralised list of files contained in the top-level
makefile.

The patches are split as follows:
1/6	The actual infrastructure added to Rules.make
2/6	makefile for atm updated
3/6	makefile for sound updated
4/6	makefile for various drivers updated
5/6	makefile in scripts/ updated
6/6	Top-level makefile updated, including removal of file-list

	Sam

[PATCH] kbuild: clean and mrproper file list created dynamically 1/6

[Sam Ravnborg]

o Create infrastructure in Rules.make to allow the individual makefiles
to specify what files shall be deleted during clean and mrproper
o Add .clean and .mrproper to ignore list

	Sam

diff -Nru a/Rules.make b/Rules.make
--- a/Rules.make	Tue Sep 10 22:37:37 2002
+++ b/Rules.make	Tue Sep 10 22:37:37 2002
@@ -405,6 +405,27 @@
 
 targets += $(host-progs-single) $(host-progs-multi-objs) $(host-progs-multi) 
 
+# Saved generated files that needs cleaning up later.
+# Filename are saved in $(objtree)/.clean and $(objtree)/.mrproper
+# When make clean or make mrproper is executed, the files listed
+# are deleted, and the file itselt is deleted.
+# To add files to the list set clean or mrproper equal to the
+# file, before inclusion of Rules.make. Rules.amke will prefix the
+# file with the full directory path.
+# ===========================================================================
+clean := $(addprefix $(CURDIR)/,$(clean))
+clean-file := $(objtree)/.clean
+tmp := $(if $(clean),\
+$(shell (if [ -f $(clean-file) ]; then cat $(clean-file); fi;\
+echo $(clean)) | sort -u > $(clean-file)))
+
+mrproper := $(addprefix $(CURDIR)/,$(mrproper) \
+$(host-progs-single) $(host-progs-multi) $(host-progs-multi-objs))
+mrproper-file := $(objtree)/.mrproper
+tmp := $(if $(mrproper),\
+$(shell (if [ -f $(mrproper-file) ]; then cat $(mrproper-file); fi;\
+echo $(mrproper)) | sort -u > $(mrproper-file)))
+
 endif # ! modules_install
 endif # ! fastdep
 

[PATCH] atm: List files to be deleted during clean and mrproper 2/6

[Sam Ravnborg]

Atm makefile updated

	Sam

diff -Nru a/drivers/atm/Makefile b/drivers/atm/Makefile
--- a/drivers/atm/Makefile	Tue Sep 10 22:37:43 2002
+++ b/drivers/atm/Makefile	Tue Sep 10 22:37:43 2002
@@ -58,6 +58,12 @@
 
 EXTRA_CFLAGS := -g
 
+# Files generated that shall be removed upon make clean
+clean := {atmsar11,pca200e,pca200e_ecd,sba200e_ecd}.{bin,bin1,bin2} 
+
+# Firmware generated that shall be removed upon make mrproper
+mrproper := fore200e_pca_fw.c fore200e_sba_fw.c
+
 include $(TOPDIR)/Rules.make
 
 # FORE Systems 200E-series firmware magic
@@ -72,6 +78,6 @@
 	  -i $(CONFIG_ATM_FORE200E_SBA_FW) -o $@
 
 # deal with the various suffixes of the binary firmware images
-$(obj)/%.bin $(obj)/%.bin1 $(obj)/%.bin2: $(obj)/%.data
+$(obj)/%.bin $(obj)/%.bin1 $(obj)/%.bin2: $(src)/%.data
 	objcopy -Iihex $< -Obinary $@.gz
 	gzip -df $@.gz

[PATCH] sound/oss: Files to be deleted during mrproper 3/6

[Sam Ravnborg]

Updates to the sound makefile

	Sam

diff -Nru a/sound/oss/Makefile b/sound/oss/Makefile
--- a/sound/oss/Makefile	Tue Sep 10 22:37:49 2002
+++ b/sound/oss/Makefile	Tue Sep 10 22:37:49 2002
@@ -96,6 +96,10 @@
 
 host-progs	:= bin2hex hex2hex
 
+# Files generated that shall be removed upon make mrproper
+mrproper := maui_boot.h msndperm.c msndinit.c pndsperm.c pndspini.c \
+pss_boot.h trix_boot.h
+
 include $(TOPDIR)/Rules.make
 
 # Firmware files that need translation

[PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6

[Sam Ravnborg]

o Specify which files to be delted during clean
	- Intermidiate files
o Specify files to be deleted during mrproper
	- Firmware

Note:The patch for the pci makefile apply with an offset of 2 lines.

	Sam

diff -Nru a/drivers/char/Makefile b/drivers/char/Makefile
--- a/drivers/char/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/char/Makefile	Tue Sep 10 22:37:55 2002
@@ -100,6 +100,9 @@
 obj-$(CONFIG_DRM) += drm/
 obj-$(CONFIG_PCMCIA) += pcmcia/
 
+# Files generated that shall be removed upon make clean
+clean := consolemap_deftbl.c defkeymap.c qtronixmap.c
+
 include $(TOPDIR)/Rules.make
 
 $(obj)/consolemap_deftbl.c: $(src)/$(FONTMAPFILE)
diff -Nru a/drivers/net/hamradio/soundmodem/Makefile b/drivers/net/hamradio/soundmodem/Makefile
--- a/drivers/net/hamradio/soundmodem/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/net/hamradio/soundmodem/Makefile	Tue Sep 10 22:37:55 2002
@@ -19,6 +19,10 @@
 host-progs := gentbl
 HOST_LOADLIBES := -lm
 
+# Files generated that shall be removed upon make clean
+clean := sm_tbl_afsk1200.h sm_tbl_afsk2400_7.h sm_tbl_afsk2400_8.h \
+sm_tbl_afsk2666.h sm_tbl_psk4800.h sm_tbl_hapn4800.h sm_tbl_fsk9600.h
+
 include $(TOPDIR)/Rules.make
 
 # Dependencies on generates files need to be listed explicitly
diff -Nru a/drivers/pci/Makefile b/drivers/pci/Makefile
--- a/drivers/pci/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/pci/Makefile	Tue Sep 10 22:37:55 2002
@@ -33,6 +33,9 @@
 
 host-progs := gen-devlist
 
+# Files generated that shall be removed upon make clean
+clean := devlist.h classlist.h
+
 include $(TOPDIR)/Rules.make
 
 # Dependencies on generated files need to be listed explicitly
diff -Nru a/drivers/scsi/Makefile b/drivers/scsi/Makefile
--- a/drivers/scsi/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/scsi/Makefile	Tue Sep 10 22:37:55 2002
@@ -129,6 +129,10 @@
 cpqfc-objs	:= cpqfcTSinit.o cpqfcTScontrol.o cpqfcTSi2c.o \
 		   cpqfcTSworker.o cpqfcTStrigger.o
 
+# Files generated that shall be removed upon make clean
+clean := 53c8xx_d.h 53c8xx_u.h 53c7xx_d.h 53c7xx_u.h \
+         sim710_d.h sim710_u.h 53c700_d.h 53c700_u.h
+
 include $(TOPDIR)/Rules.make
 
 $(obj)/53c7,8xx.o: $(obj)/53c8xx_d.h $(obj)/53c8xx_u.h
@@ -159,4 +163,4 @@
 $(obj)/53c700_d.h: $(src)/53c700.scr $(src)/script_asm.pl
 	$(PERL) -s $(src)/script_asm.pl -ncr7x0_family $@ $(@:_d.h=_u.h) < $<
 
-endif
\ No newline at end of file
+endif
diff -Nru a/drivers/scsi/aic7xxx/Makefile b/drivers/scsi/aic7xxx/Makefile
--- a/drivers/scsi/aic7xxx/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/scsi/aic7xxx/Makefile	Tue Sep 10 22:37:55 2002
@@ -20,6 +20,14 @@
 
 #EXTRA_CFLAGS += -g
 
+# Files generated that shall be removed upon make clean
+clean := aic7xxx_seq.h aic7xxx_reg.h
+
+# Files generated that shall be removed upon make mrproper
+# The list includes all files listed for the clean: target in aicasm/Makefile
+mrproper := $(addprefix \
+aicasm/,aicasm_gram.c aicasm_scan.c y.tab.h aicdb.h y.output aicasm)
+
 include $(TOPDIR)/Rules.make
 
 # Dependencies for generated files need to be listed explicitly
diff -Nru a/drivers/video/Makefile b/drivers/video/Makefile
--- a/drivers/video/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/video/Makefile	Tue Sep 10 22:37:55 2002
@@ -118,6 +118,9 @@
 obj-$(CONFIG_FBCON_STI)           += fbcon-sti.o
 obj-$(CONFIG_FBCON_ACCEL)	  += fbcon-accel.o
 
+# Files generated that shall be removed upon make clean
+clean := promcon_tbl.c
+
 include $(TOPDIR)/Rules.make
 
 $(obj)/promcon_tbl.c: $(src)/prom.uni
diff -Nru a/drivers/zorro/Makefile b/drivers/zorro/Makefile
--- a/drivers/zorro/Makefile	Tue Sep 10 22:37:55 2002
+++ b/drivers/zorro/Makefile	Tue Sep 10 22:37:55 2002
@@ -9,6 +9,9 @@
 
 host-progs 		:= gen-devlist
 
+# Files generated that shall be removed upon make clean
+clean := devlist.h
+
 include $(TOPDIR)/Rules.make
 
 # Dependencies on generated files need to be listed explicitly

[PATCH] scripts: Removed mrproper targets, they are now handled automagically 5/6

[Sam Ravnborg]

Removed mrproper targets from scripts/Makefile and scripts/lxdialog/Makefile
The new clean: and mrproper: handling will delete the files in question.

For lxdialog added a few files that could be left if build were stopped
unexpected.

	Sam

diff -Nru a/scripts/Makefile b/scripts/Makefile
--- a/scripts/Makefile	Tue Sep 10 22:38:01 2002
+++ b/scripts/Makefile	Tue Sep 10 22:38:01 2002
@@ -9,6 +9,8 @@
 # conmakehash:	 Create arrays for initializing the kernel console tables
 # tkparse: 	 Used by xconfig
 
+clean    := kconfig.tk
+
 all: fixdep split-include docproc conmakehash __chmod
 
 # The following temporary rule will make sure that people's
@@ -57,9 +59,4 @@
 $(obj)/split-include $(obj)/docproc $(addprefix $(obj)/,$(tkparse-objs)) \
 $(obj)/conmakehash lxdialog: $(obj)/fixdep
 
-mrproper:
-	@echo 'Making mrproper (scripts)'
-	@rm -f $(tkparse-objs) $(obj)/kconfig.tk
-	@rm -f core $(host-progs)
-	@$(MAKE) -C lxdialog mrproper
 
diff -Nru a/scripts/lxdialog/Makefile b/scripts/lxdialog/Makefile
--- a/scripts/lxdialog/Makefile	Tue Sep 10 22:38:01 2002
+++ b/scripts/lxdialog/Makefile	Tue Sep 10 22:38:01 2002
@@ -15,8 +15,12 @@
 endif
 endif
 
-host-progs := lxdialog
+# Files removed during make clean
+clean := a.out lxtemp.c
 
+# Host executable
+host-progs := lxdialog
+# Additional object files for lxdialog
 lxdialog-objs := checklist.o menubox.o textbox.o yesno.o inputbox.o \
 		 util.o lxdialog.o msgbox.o
 
@@ -39,5 +43,3 @@
 		exit 1 ;\
 	fi
 
-mrproper:
-	@rm -f core $(host-progs) $(lxdialog-objs) ncurses

[PATCH] Distibuted clean and mrproper handling 6/6

[Sam Ravnborg]

The makefiles that know what files to be deleted does now 
specify them using
clean := files
and
mrproper := files

Additionally all programs build using host-progs are deleted as well.
This allowed us to remove a central list of files in the top-level makefile,
and by moving the responsibility out to the makefiles that has the knowledge
there is a good chance the information will stay updated.

	Sam

diff -Nru a/Makefile b/Makefile
--- a/Makefile	Tue Sep 10 22:38:08 2002
+++ b/Makefile	Tue Sep 10 22:38:08 2002
@@ -604,41 +604,11 @@
 CLEAN_FILES += \
 	include/linux/compile.h \
 	vmlinux System.map \
-	drivers/char/consolemap_deftbl.c drivers/video/promcon_tbl.c \
-	drivers/char/conmakehash \
-	drivers/char/drm/*-mod.c \
-	drivers/char/defkeymap.c drivers/char/qtronixmap.c \
-	drivers/pci/devlist.h drivers/pci/classlist.h drivers/pci/gen-devlist \
-	drivers/zorro/devlist.h drivers/zorro/gen-devlist \
-	sound/oss/bin2hex sound/oss/hex2hex \
-	drivers/atm/fore200e_mkfirm drivers/atm/{pca,sba}*{.bin,.bin1,.bin2} \
-	drivers/scsi/aic7xxx/aic7xxx_seq.h \
-	drivers/scsi/aic7xxx/aic7xxx_reg.h \
-	drivers/scsi/aic7xxx/aicasm/aicasm_gram.c \
-	drivers/scsi/aic7xxx/aicasm/aicasm_scan.c \
-	drivers/scsi/aic7xxx/aicasm/y.tab.h \
-	drivers/scsi/aic7xxx/aicasm/aicasm \
-	drivers/scsi/53c700_d.h drivers/scsi/sim710_d.h \
-	drivers/scsi/53c7xx_d.h drivers/scsi/53c7xx_u.h \
-	drivers/scsi/53c8xx_d.h drivers/scsi/53c8xx_u.h \
-	net/802/cl2llc.c net/802/transit/pdutr.h net/802/transit/timertr.h \
-	net/802/pseudo/pseudocode.h \
-	net/khttpd/make_times_h net/khttpd/times.h \
 	submenu*
 
 # 	files removed with 'make mrproper'
 MRPROPER_FILES += \
 	include/linux/autoconf.h include/linux/version.h \
-	drivers/net/hamradio/soundmodem/sm_tbl_{afsk1200,afsk2666,fsk9600}.h \
-	drivers/net/hamradio/soundmodem/sm_tbl_{hapn4800,psk4800}.h \
-	drivers/net/hamradio/soundmodem/sm_tbl_{afsk2400_7,afsk2400_8}.h \
-	drivers/net/hamradio/soundmodem/gentbl \
-	sound/oss/*_boot.h sound/oss/.*.boot \
-	sound/oss/msndinit.c \
-	sound/oss/msndperm.c \
-	sound/oss/pndsperm.c \
-	sound/oss/pndspini.c \
-	drivers/atm/fore200e_*_fw.c drivers/atm/.fore200e_*.fw \
 	.version .config* config.in config.old \
 	.menuconfig.log \
 	include/asm \
@@ -658,6 +628,8 @@
 
 clean:	archclean
 	@echo 'Cleaning up'
+	@if [ -f $(obj)/.$@ ]; then rm -f `cat $(obj)/.$@`; fi;
+	@rm -f $(obj)/.$@
 	@find . -name SCCS -prune -o -name BitKeeper -prune -o \
 		\( -name \*.[oas] -o -name core -o -name .\*.cmd -o \
 		-name .\*.tmp -o -name .\*.d \) -type f -print \
@@ -667,12 +639,13 @@
 
 mrproper: clean archmrproper
 	@echo 'Making mrproper'
+	@if [ -f $(obj)/.$@ ]; then rm -f `cat $(obj)/.$@`; fi;
+	@rm -f .$@
 	@find . -name SCCS -prune -o -name BitKeeper -prune -o \
 		\( -name .depend -o -name .\*.cmd \) \
 		-type f -print | xargs rm -f
 	@rm -f $(MRPROPER_FILES)
 	@rm -rf $(MRPROPER_DIRS)
-	@$(MAKE) -C scripts mrproper
 	@$(MAKE) -f Documentation/DocBook/Makefile mrproper
 
 distclean: mrproper

Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6

[Justin T. Gibbs]

--On Tuesday, September 10, 2002 23:06:56 +0200 Sam Ravnborg
<sam@ravnborg.org> wrote:
 ..

> diff -Nru a/drivers/scsi/aic7xxx/Makefile b/drivers/scsi/aic7xxx/Makefile
> --- a/drivers/scsi/aic7xxx/Makefile	Tue Sep 10 22:37:55 2002
> +++ b/drivers/scsi/aic7xxx/Makefile	Tue Sep 10 22:37:55 2002
> @@ -20,6 +20,14 @@
>  
>  #EXTRA_CFLAGS += -g
>  
> +# Files generated that shall be removed upon make clean
> +clean := aic7xxx_seq.h aic7xxx_reg.h

At lease this line need to be contingent on the actual building of
firmware.  Otherwise you've just blown away the firmware the vendor
has shipped with the system and the user may not have the utilities
to rebuild it.

--
Justin

Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6

[Sam Ravnborg]

On Tue, Sep 10, 2002 at 07:07:58PM -0600, Justin T. Gibbs wrote:
> > +# Files generated that shall be removed upon make clean
> > +clean := aic7xxx_seq.h aic7xxx_reg.h
> 
> At lease this line need to be contingent on the actual building of
> firmware.  Otherwise you've just blown away the firmware the vendor
> has shipped with the system and the user may not have the utilities
> to rebuild it.
The original firmware are stored in files named:
aic7xxx_reg.h_shipped  aic7xxx_seq.h_shipped
They are copied to aic7xxx_seq.h aic7xxx_reg.h during the build.
So no problem here.

Snip from original Makefile:
CLEAN_FILES :=
...
        drivers/scsi/aic7xxx/aic7xxx_seq.h \
        drivers/scsi/aic7xxx/aic7xxx_reg.h \
        drivers/scsi/aic7xxx/aicasm/aicasm_gram.c \
        drivers/scsi/aic7xxx/aicasm/aicasm_scan.c \
        drivers/scsi/aic7xxx/aicasm/y.tab.h \
        drivers/scsi/aic7xxx/aicasm/aicasm \

As you can see the files in question has always been deleted during
a make clean.

The only functional change done is the fact that firmware are now deleted
by make mrproper, as is true for rest of the kernel.

	Sam

Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6

[Andrew Bray]

On Wed, 11 Sep 2002 07:12:19 +0200, Sam Ravnborg <sam@ravnborg.org> wrote:
>On Tue, Sep 10, 2002 at 07:07:58PM -0600, Justin T. Gibbs wrote:
>> > +# Files generated that shall be removed upon make clean
>> > +clean := aic7xxx_seq.h aic7xxx_reg.h
>> 
>> At lease this line need to be contingent on the actual building of
>> firmware.  Otherwise you've just blown away the firmware the vendor
>> has shipped with the system and the user may not have the utilities
>> to rebuild it.
>The original firmware are stored in files named:
>aic7xxx_reg.h_shipped  aic7xxx_seq.h_shipped
>They are copied to aic7xxx_seq.h aic7xxx_reg.h during the build.
>So no problem here.

Just a thought: should 'make clean' and/or 'make mrproper' rename the
_shipped files back to their origilal names?

Regards,

Andy

-- 
-----------------------------------------------------------------------------
Andrew Bray, PWMS, MA,              |  preferred:    mailto:andy@chaos.org.uk
London, England                     |  or:   mailto:andy@madhouse.demon.co.uk
PGP id/fingerprint:  D811F5C9/26 B5 42 C6 F4 00 B2 71 BA EA 9B 81 6C 65 59 07

ADMIN: DON'T try to be clever with email headers!

[Matti Aarnio]

  Folks, when aiming to post into VGER's lists, DO NOT
  try to make any cute things in headers.  Any such
  are bound to cause TONS of bounces, which did happen
  in this particular case...

  This is particularly important when posting patches..


On Fri, Sep 13, 2002 at 03:49:01PM +0000, Andrew Bray wrote:
> To:	<linux-kernel@vger.kernel.org>
> From:	abuse@madhouse.demon.co.uk (Andrew Bray)
> Subject: Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6
> Date:	13 Sep 2002 15:49:01 GMT
> Organization: Private Internet Node
> Message-ID: <slrnao427d.si3.abuse@madhouse.demon.co.uk>
> Reply-To:	andy@@chaos.org.uk

Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6

[Sam Ravnborg]

On Fri, Sep 13, 2002 at 03:49:01PM +0000, Andrew Bray wrote:
> Just a thought: should 'make clean' and/or 'make mrproper' rename the
> _shipped files back to their origilal names?
What could be done is to delete the files generated from _shipped.
Hereby they would be generated from their _shipped variants during next
the next build. Please note that they are not renamed, but copied.
But I wanted to keep current functionality and later extend as
made sense. A stepwise approach.

I also had in mind only deleting .o files generated during the build,
allowing us to get rid of the find ... in make clean.
make mrproper would need it to remove stale .o files, and the like.

Another item that popped up while doing this was a new definition of
mrproper and clean.
They are clearly mixed, and it seems mrproper is used to clean everything
when kbuild is confused. But very often people want to keep their
.config. Dunno yet what to do, and for sure it makes no sense whatsoever
to mix a clean-up patch with a new functionality patch.

	Sam

Re: ADMIN: DON'T try to be clever with email headers!

[Andrew Bray]

-----BEGIN PGP SIGNED MESSAGE-----

Hi Matti,

On Fri, 13 Sep 2002, Matti Aarnio wrote:

>   Folks, when aiming to post into VGER's lists, DO NOT
>   try to make any cute things in headers.  Any such
>   are bound to cause TONS of bounces, which did happen
>   in this particular case...
> 
>   This is particularly important when posting patches..
> 
> 
> On Fri, Sep 13, 2002 at 03:49:01PM +0000, Andrew Bray wrote:
> > To:	<linux-kernel@vger.kernel.org>
> > From:	abuse@madhouse.demon.co.uk (Andrew Bray)
> > Subject: Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6
> > Date:	13 Sep 2002 15:49:01 GMT
> > Organization: Private Internet Node
> > Message-ID: <slrnao427d.si3.abuse@madhouse.demon.co.uk>
> > Reply-To:	andy@@chaos.org.uk

I object!

There is nothing 'cute' in this header, and I didn't post a patch, just a
comment on one.  It was entirely correct SMTP, with a truly unique message
ID and all addresses were valid for sending replies (From:, and
Reply-To:), and all arrive to me.

The SMTP envelope from address (which you didn't include) is another valid 
address which I don't expect to cause trouble: mail-list@madhouse).

abuse@madhouse is a perfectly valid e-mail address on this machine, and 
andy@chaos is an address which will always reach me, even if I leave 
demon.

I chose abuse as an address, because I have received many spam messages 
offering 'cleaned' address lists where obvious 'trap-door' addresses were 
filtered out - 'abuse@' was explicitely mentioned.  This I count as a 
success, because, despite using abuse@ in the From: address in all news 
and mailing list postings for six years, only 14% of my incoming spam is 
sent to abuse@madhouse. (Interestingly, only 1% of incoming spam is sent 
to my Reply-To: address).

I cannot imagine why folks would want to trap abuse as a From: address, 
because in the thousands of spam messagess I have received over the last 
ten years, NONE have had abuse in either the envelope from address, or the 
SMTP From: header.

abuse is traditionally a destination address for reporting abuse to ISPs, 
for example demon have an email address for reporting abuse by customers 
which is abuse@demon.co.uk

I argue that the bounces are coming from mis-configured MTAs, as:

1) They should be bouncing on the envelope-from address rather than the 
   From: header, as the envelope address is more likely to be valid.  MTAs 
   shoud regard the From: address as a comment, because it is so readily 
   forged.

2) abuse is a silly address to trap, as it just doesn't happen in spam.

I use a threading news reader to read linux-kernel, and the scripts I use 
can easily munge the From: address to point back to the processor (I 
belong to some mailing lists that only accept postings from the address 
that is registered), but then then the mail would have a From: address 
that isn't read by a human, which in my opinion is real abuse.

This mailing list is gatewayed to a number if widely available usenet 
newsgroups, so I am loath to return to using my 'real' address in the 
From: header.

Regards,

Andy

- -- 
- -----------------------------------------------------------------------------
Andrew Bray, PWMS, MA,              |  preferred:    mailto:andy@chaos.org.uk
London, England                     |  or:   mailto:andy@madhouse.demon.co.uk
PGP id/fingerprint:  D811F5C9/26 B5 42 C6 F4 00 B2 71 BA EA 9B 81 6C 65 59 07

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2i

iQCVAwUBPYJx0kjMoRfYEfXJAQFV1AP8DPA4kWPQ3xHYMiRK7c4vdhRg9DzKt/TP
yuEIZQkYBv1F+Lhv7kuNsQ5f8VkREYF0MvBih+liMIRIRBZqbxofwnm/59+rz/aZ
GrpwSPXm3cCbkC9eyuTDTlTcdNmoGk5sVwyirLBYHb4mr8mvpJM9dE6ND2KoaLjS
gTEqNIjjfxU=
=fZlb
-----END PGP SIGNATURE-----

Re: ADMIN: DON'T try to be clever with email headers!

[Matti Aarnio]

On Sat, Sep 14, 2002 at 12:16:15AM +0100, Andrew Bray wrote:
> Hi Matti,
...
> > On Fri, Sep 13, 2002 at 03:49:01PM +0000, Andrew Bray wrote:
> > > To:	<linux-kernel@vger.kernel.org>
> > > From:	abuse@madhouse.demon.co.uk (Andrew Bray)
> > > Subject: Re: [PATCH] drivers/pci,hamradio,scsi,aic7xxx,video,zorro clean and mrproper files 4/6
> > > Date:	13 Sep 2002 15:49:01 GMT
> > > Organization: Private Internet Node
> > > Message-ID: <slrnao427d.si3.abuse@madhouse.demon.co.uk>
> > > Reply-To:	andy@@chaos.org.uk
> 
> I object!
> 
> There is nothing 'cute' in this header, and I didn't post a patch, just a
> comment on one.  It was entirely correct SMTP, with a truly unique message
> ID and all addresses were valid for sending replies (From:, and
> Reply-To:), and all arrive to me.

  Look closer at the "Reply-To:" header.  THAT caused tons of bounces.

  You didn't post a patch, but there has been recently quite a lot of
  talk about patches not reaching people when they filter things, and
  throw messages with even slightly suspicuous headers into dev-null..

....  (lots of semi-relevant text (which I even agree with) dropped ..)

  I really would love to have a globally functional email system
  where everybody sends PGP encrypted messages over SSL encrypted
  email connections, everybody posts to LIST public key, list opens
  it up, and posts to each recipient's public key..
  TONS of encryption work at the list processors, of course..

  How can that prevent the list from making into spammers address
  harvesters ?  Not anymore than the current situation.  Any list
  archive with open access is prone to reveal the addresses.

> Regards,
> Andy
>----------------------------------------------------------------------------
> Andrew Bray, PWMS, MA,              |  preferred:    mailto:andy@chaos.org.uk
> London, England                     |  or:   mailto:andy@madhouse.demon.co.uk

Re: ADMIN: DON'T try to be clever with email headers!

[Matthias Andree]

On Fri, 13 Sep 2002, Matti Aarnio wrote:

>   Folks, when aiming to post into VGER's lists, DO NOT
>   try to make any cute things in headers.  Any such
>   are bound to cause TONS of bounces, which did happen
>   in this particular case...

The fix is to normalize headers. Any mailing list should do that. I have
seen ISODE PP 5.0, running the uni-dortmund secondary MX (which is in
fact the only reachable outside the intranet), bounce on perfectly-legal
addresses, and mails containing double Date: headers and for other
frivolous reasons. Recently I got unsubscribed for reasons Dave Miller
could not recall, he had not kept the bounce :-(

Does anybody know of an opensource real-time anti-relay SMTP proxy that
can do aliasing?

I like the way ezmlm-idx, a mailing list manager, handles things, it
stores bounces, and after some days, notifies the list members, and if
the notice does not get through, kills the subscription after another
verification. I'd be good if the vger lists could do the same, so users
could actually figure what goes through and what is rejected.

Anyhow, by all means, DO KEEP THE BOUNCES YOU GET, use a script and keep
them for two weeks, and then weed them out. Feel free to weed all
subsequent bounces from one address that has the same bounce reason.
This should be possible with low effort, but allows users to figure
what's up should they ask.

BTW, his header showed up escaped at my site:

Reply-To: "andy@"@his.domain

And, FWIW, this looks legal to me according to RFC-2822 EBNF.