On 11/11/06, Anselm R. Garbe <[EMAIL PROTECTED]> wrote:
----- Forwarded message from [EMAIL PROTECTED] -----
Date: Sat, 11 Nov 2006 11:52:32 +0100
From: [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Subject: Auto-discard notification
The attached message has been automatically discarded.
Date: Sat, 11 Nov 2006 10:52:31 +0000
From: Salva Peiró <[EMAIL PROTECTED]>
To: [email protected]
Subject: a small improvement
Hi,
this is a silly thing, but i thought it was worth mentioning:
diff -r a22a70174d44 Makefile
--- a/Makefile Wed Nov 8 16:16:38 2006
+++ b/Makefile Sat Nov 11 11:30:06 2006
@@ -21,9 +21,9 @@
${OBJ}: dwm.h config.h config.mk
-config.h:
- @echo creating $@ from config.default.h
- @cp config.default.h $@
+config.h: config.gmt.h
+ @echo creating $@ from $<
+ @cp $< $@
dwm: ${OBJ}
@echo LD $@
I would do the creation of the config.h header this way, because
i usually store my config in a separate (config.gmt.h) file, and when
a change is done on default config.h's provided i take a look at them
and merge changes on my own config, this way just after having modified
the config.gmt.h issuing `make' updates config.h and builds an up to date
dwm.
I don't think that's a good idea. The reason config.h has no
dependencies is that it should *never* be overwritten if it already
exists. The idea behind the current structure was to allow people to
keep their custom configuration in config.h, guaranteeing that it'll
never be overwritten by the build, while still making a simple 'make
install' work even when no custom config.h is present (by creating one
from the default). The most important part here though was that
config.h should *always* remain intact if it already exists. This has
been discussed at quite some length a couple of months ago (arg: maybe
it's a good idea to make some kind of FAQ about such design
decisions?).
Greets, Sander.
