18.04.2014 15:55, Stefan Weil wrote: > Use an absolute value for SRC_PATH. This fixes a build problem: > > $ LANG=C make -C po update > make: Entering directory `/qemu/po' > GEN ../po/messages.po > /bin/sh: 1: cannot create ../po/messages.po: Directory nonexistent > make: *** [../po/messages.po] Error 2 > make: Leaving directory `/qemu/po' > > Signed-off-by: Stefan Weil <s...@weilnetz.de> > --- > po/Makefile | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/po/Makefile b/po/Makefile > index 705166e..576b172 100644 > --- a/po/Makefile > +++ b/po/Makefile > @@ -2,7 +2,7 @@ > # process and also within the source tree to update the translation files. > > # Set SRC_PATH for in-tree builds without configuration. > -SRC_PATH=.. > +SRC_PATH=$(shell cd .. && pwd) > > -include ../config-host.mak > include $(SRC_PATH)/rules.mak
How about moving it below inclusion of config-host.mak and using something like SRC_PATH ?= $(realpath ..) ? I dunno how gnu'ish this construct is... :) However, I think I've a better fix for this: Author: Michael Tokarev <m...@tls.msk.ru> Date: Sun Apr 27 13:32:07 2014 +0400 po/Makefile: fix $SRC_PATH reference The rule for messages.po appears to be slightly wrong. Move the `cd' command within parens. Signed-off-by: Michael Tokarev <m...@tls.msk.ru> Cc: Stefan Weil <s...@weilnetz.de> Signed-off-by: Michael Tokarev <m...@tls.msk.ru> diff --git a/po/Makefile b/po/Makefile index 705166e..669f865 100644 --- a/po/Makefile +++ b/po/Makefile @@ -37,8 +37,8 @@ install: $(OBJS) $(call quiet-command, msgfmt -o $@ $<, " GEN $@") $(PO_PATH)/messages.po: $(SRC_PATH)/ui/gtk.c - $(call quiet-command, cd $(SRC_PATH) && \ - (xgettext -o - --from-code=UTF-8 --foreign-user \ + $(call quiet-command, ( cd $(SRC_PATH) && \ + xgettext -o - --from-code=UTF-8 --foreign-user \ --package-name=QEMU --package-version=$(VERSION) \ --msgid-bugs-address=qemu-devel@nongnu.org -k_ -C ui/gtk.c | \ sed -e s/CHARSET/UTF-8/) >$@, " GEN $@") Note the already existing parens () -- it looks like this is in order to run commands within a subshell (with its own current directory) but still have correct output redirection. If you don't have objections, I'll apply my version (it also fixes the issue). Thanks, /mjt Thanks, /mjt