Note that if you just want to use gale to chat, and not run a server, there are easier ways. such as http://yammer.net/ and standalone clients like ginsu
Download the latest version from http://download.ofb.net/gale/ and unpack it. It should create a directory named "gale-(version)", where (version) is the version number you downloaded. (We'll refer to this directory as "gale".)
Get ready to run the provided "configure" script. The typical "configure" options (--prefix etc.) apply; run "sh configure --help" for usage information.
Include the appropriate -L and -I flags in the LDFLAGS and CPPFLAGS environment variables (respectively) so "configure" can find libcrypto.a (or libcrypto.so) and the OpenSSL? headers (openssl/evp.h and friends).
Similarly, make sure $LDFLAGS and $CPPFLAGS include the location of libgc.a (or libgc.so) and gc.h, respectively. ADNS, libiconv, and any other required libraries must also be available via these paths.
By default and where supported, Gale will install static and shared libraries, and install binaries linked against shared libraries. You may use --disable-shared or --disable-static to modify this behavior.
Once you've figured out the right options and set the right environment variables, run the script.
Run "make" in the top-level Gale directory. With luck, this will build everything without incident. If not, now you get to figure out what went wrong and why; enjoy! Feel free to contact me (egnor-gale at ofb.net) with questions and patches. Many architectures are currently unsupported, but the changes are typically not too painful.
Then become a user (eg. "root") which has permission to write (and create, if necessary) the install directories (/usr/local/..., unless you specified otherwise). Run "make install" from the top-level "gale" directory; this should install the Gale binaries, libraries, and header files.
Make sure the installed binaries are on the PATH for future steps!
Fifth, run the "gale-install" script that was installed by "make install". This may ask questions if it can't find a previous configuration. Feel free to interrupt and re-start it at any point.
Finally, if you have not already set up your domain, you need to do so. Become the "gale user" selected in the last step and run "gale-install" again. Follow the directions; these will involve sending your key to the proper authority to get it signed. I manage ROOT and most "top-level" domains (.edu, .com, .org, .nu, .ml.org, etc.) for now. When in doubt, ask me.
Re-run "gale-install" as the gale user. It will ask you for the file containing the signed key. It will then take that and install it in the correct place for you and complete the setup.
Enjoy your new installation. If you have any problems, feel free to send e-mail to egnor-gale at ofb.net.
The stock source code assumes iconv is in its own library, and furthermore that unsigned -> signed assignment doesn't generate a warning. This patch will let you use iconv with modern glibcs and gccs.
![[Home]](/usemod-wiki/debian-logo.png){kind=logo}
--- gale-0.99fruit-orig/configure 2004-02-19 21:13:59.000000000 +0000
+++ gale-0.99fruit/configure 2006-11-29 21:56:15.582370000 +0000
@@ -7747,7 +7747,6 @@
echo $ECHO_N "checking for working iconv... $ECHO_C" >&6
save_libs="$LIBS"
save_flags="$CFLAGS"
-LIBS="$LIBS -liconv"
CFLAGS="$error_flags $save_flags -DICONV_CONST="
if test "$cross_compiling" = yes; then
echo "$as_me:$LINENO: result: unknown, assuming no" >&5
--- gale-0.99fruit-orig/iconvtest.c 2000-11-18 06:38:47.000000000 +0000
+++ gale-0.99fruit/iconvtest.c 2006-11-29 21:53:05.945193000 +0000
@@ -27,7 +27,7 @@
|| utf[0] != 0x7f
|| utf[1] != 0xdf || utf[2] != 0xbf
|| utf[3] != 0xef || utf[4] != 0xbf || utf[5] != 0xbf) exit(1);
- inbuf = utf; inbytes = 6;
+ inbuf = (char *) utf; inbytes = 6;
outbuf = (char *) wch; outbytes = sizeof(wch);
if (0 != iconv(from,&inbuf,&inbytes,&outbuf,&outbytes)
|| 0 != inbytes || (sizeof(wch) - 3*sizeof(wchar_t)) != outbytes
Note that it may be necessary to add "-ldl" to the LDFLAGS environment variable when running "configure" on the main gale source, in order to overcome an apparent bug.
The RPMs at http://www.slackers.net/gale/ will work on Redhat 7.3, but not 9.0 (as of 11 Nov 2003). Redhat 9.0 users should upgrade to at least GC 6.2.
I recently compiled from source using the above instructions, and found them incomplete, so below is a detailed description of what I did. This is not the only way to do so, and may not even be the best way.
-- Beland
I got these files:
gale.tar.gz from http://gale.org/download gc6.2.tar.gz from http://www.hpl.hp.com/personal/Hans_Boehm/gc/gc_source/ openssl-devel-0.9.7a-20.i386.rpm from the Redhat distribution (see http://rpmfind.net/) adns-0.9-2.i686.rpm from http://www.slackers.net/gale/ (I did not bother with iconv)
I installed the relevant RPMs with:
rpm -Uvh openssl-devel-0.9.7a-20.i386.rpm rpm -Uvh adns-0.9-2.i686.rpm
In the directory I downloaded gale.tar.gz and gc6.2.tar.gz, I ran the following commands:
gtar -xzvf gale.tar.gz gtar -xzvf gc6.2.tar.gz
cd gc6.2/
./configure --prefix=/usr/local/gc
make
("su" to become root)
make install
(add the line "/usr/local/gc/lib" to /etc/ld.so.conf)
("exit" to become a normal user again)
This makes, among other files:
/usr/local/gc/lib/libgc.a /usr/local/gc/include/gc.h
Then I ran these commands:
cd ../gale-0.99fruit/
setenv LDFLAGS "-ldl -L/usr/local/gc/lib"
setenv CPPFLAGS -I/usr/local/gc/include
(run "tcsh" if the above 2 lines don't work)
./configure --prefix=/usr/local/gale
make
("su" to become root)
make install
Now all the Gale binaries are in /usr/local/gale/bin/, so I do the following before running anything (easy enough to put in my .cshrc):
setenv PATH `printenv PATH`:/usr/local/gale/bin
Be sure to run:
/usr/local/gale/bin/gale-install