[Acl-devel] Compilation question

[Paul Rogers]

Hi,

First time poster so apologies if I miss out on any protocols.

I'm having a problem compiling acl-2.2.52 and am wondering if there's a 
bug in .... Here's my setup:

* Why am I compiling? I'm running on a Synology NAS so cannot just get 
the RPM. Plus I need libacl for compiling Rsync with ACL support as the 
files to be stored on the Synology from a remote NAS have extended ACLs.
* attr v2.4.47 is compiled by me and installed in to /opt/local/attr
* Include files from attr are symlinked into /opt/local/include (sourced 
from /opt/local/attr/include).
* Libs from attr are symlinked in a similar manner into /opt/local/lib 
(sourced from /opt/local/attr/lib).

When trying to compile acl-2.2.52, I get an error that xattr.h cannot be 
found. Here's what I ran (have trimmed the output):

$ LDFLAGS=-L/opt/local/lib CFLAGS=-I/opt/local/include 
CPPFLAGS=-I/opt/local/include ./configure --prefix=/opt/local/acl
$ make
/opt/bin/make -C libacl
make[2]: Entering directory `/tmp/acl-2.2.52/libacl'
/opt/bin/bash ../libtool --mode=compile gcc -g -O2 -DDEBUG -funsigned-char 
-fno-strict-aliasing -Wall -DVERSION=\"2.2.52\" 
-DLOCALEDIR=\"/opt/local/acl/share/locale\" -DPACKAGE=\"acl\" 
-I../include -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -include perm_copy.h  -c 
acl_delete_def_file.c
libtool: compile:  gcc -g -O2 -DDEBUG -funsigned-char -fno-strict-aliasing 
-Wall -DVERSION=\"2.2.52\" 
-DLOCALEDIR=\"/opt/local/acl/share/locale\" -DPACKAGE=\"acl\" 
-I../include -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -include perm_copy.h -c 
acl_delete_def_file.c  -fPIC -DPIC -o .libs/acl_delete_def_file.o
acl_delete_def_file.c:23:24: fatal error: attr/xattr.h: No such file or 
directory

Question - should CFLAGS from configure be honoured in the compilation 
process (i.e. adding it to GCFLAGS in include/builddefs)? Reason I ask 
that is gcc is not being told to look for xattr.h if it's in a 
non-standard location. By adding -I/opt/local/include to GCFLAGS in 
include/builddefs means libacl compiles.

If someone who knows more about this can confirm this is a problem, am 
happy to raise a bug report. Didn't want to do that in case I am missing 
something.

Thanks in advance,

Paul.