Gnome's bugzilla was refusing attachments so forgive me for the inline  
comment.  
  
This patch fixes the comparison logic when testing the gpg version.  
  
It also applies the same logic for gpgme cause I put the code in the wrong  
spot originally and saw no need to remove it.  
--- configure.in.orig   2004-11-01 03:24:04.000000000 +0930  
+++ configure.in        2004-12-19 21:11:35.832207728 +0930  
@@ -52,13 +52,17 @@  
        micro=`echo $gnupg_version | \  
                sed 's/^gpg (GnuPG) \([[0-9]]*\)\.\([[0-9]]*\)\. 
\([[0-9]]*\).*/\3/'`  
  
-         if test "$major" -eq "$req_major"; then  
-            if test "$minor" -eq "$req_minor"; then  
-               if test "$micro" -ge "$req_micro"; then  
-                      ok="yes"  
-               fi  
+       if test "$major" -gt "$req_major"; then  
+           ok="yes"  
+        elif test "$major" -eq "$req_major"; then  
+          if test "$minor" -gt "$req_minor"; then  
+               ok="yes"  
+          elif test "$minor" -eq "$req_minor"; then  
+            if test "$micro" -ge "$req_micro"; then  
+              ok="yes"  
             fi  
-         fi  
+          fi  
+       fi  
   fi  
  
   if test "$ok" = "yes"; then  
@@ -93,8 +97,12 @@  
       micro=`echo $gpgme_config_version | \  
                 sed 's/\([[0-9]]*\)\.\([[0-9]]*\)\.\([[0-9]]*\).*/\3/'`  
  
-       if test "$major" -eq "$req_major"; then  
-          if test "$minor" -eq "$req_minor"; then  
+       if test "$major" -gt "$req_major"; then  
+           ok="yes"  
+        elif test "$major" -eq "$req_major"; then  
+          if test "$minor" -gt "$req_minor"; then  
+               ok="yes"  
+          elif test "$minor" -eq "$req_minor"; then  
             if test "$micro" -ge "$req_micro"; then  
               ok="yes"  
             fi  
  

Thanks for the patch. However, all seahorse functionality that is directly
dependant on GPG needs to be tested with the new versions of GPG, to ensure
compatibility. This includes:

- Modifying of gpg.conf when using seahorse-agent
- GPG keyserver plugins (only found in CVS)
- Exporting of secret keys (soon to arrive in CVS)

The current method for using unstable GPG versions with seahorse is to pass
--disable-gpg-check to ./configure, in which case you assume the risks of using
an untested version.

Please reopen this bug if this doesn't work for you. 

BTW, the message (install version 1.2.1 or greater) has been clarified in the
CVS version of seahorse. 

Hmmm, I just saw that GPG has in fact released 1.4 as stable several days ago.
This means we're going to have to get the two working together. I'll reopen this
bug. 

Since we made sure seahorse-agent works correctly with the new version, and I
will vouch that searching LDAP keyservers still works, what else needs to be
checked before we edit the configure script?

As outlined on seahorse-devel we need to come up with a list of ways we interact
with GPG directly, then we can check those.

I've committed changes which allow use with GPG 1.4. Adam, could you check if it
works properly, and also look at the list in the HACKING file of the direct uses
of GPG to see if they all still work properly. 

As nearly as I can tell, everything listed in HACKING works as specified with
GPG 1.4.0.  The only suggestion I can make and this is not pertaining to
operation is when seahorse-agent adds its agent string to the bottom of
gpg.conf, it doens't use double slashes.

*  gpg-agent-info /tmp//seahorse-20aJT7/S.gpg-agent:7149:1
*  use-agent

This doesn't affect usability or functionality, it just looks wrong.

Thanks for testing. The _PATH_TMP problem is one that has to actually be coded
in, as some platforms put a trailing slash on _PATH_TMP and others don't. Don't
have time to fix that today before the release. But I guess we can get to it later.