We looked closer at the code and decided that it was actually safe to just get
rid of the cast altogether, so the configure test is actually no longer
needed. Simply grab SVN HEAD and things should work.
On Wednesday, August 18, 2010 10:07:55 pm Brian Gough wrote:
> At Fri, 13 Aug 2010 18:41:46 +0200,
> Christian Grothoff wrote:
> > Now, I don't know when exactly this type was introduced, but it is
> > clearly the preferred form at this point. I've also not hit an sqlite3
> > installation that does not have it -- could you double-check that the
> > compiler is not picking up some "older" version of the header somehow?
> Hello.. I've checked the installed version (sqlite 3.4.2) and the
> sqlite3_int64 definition is not present in that version, they must
> have changed it after that.
> (Fencepost is using Ubuntu 8.04 LTS, which is old but fairly common.)
> As an alternative, here's a patch which tests for sqlite3_int64 in the
> configure script and uses the appropriate definition otherwise.
> Index: src/datastore/plugin_datastore_sqlite.c
> --- src/datastore/plugin_datastore_sqlite.c (revision 12650)
> +++ src/datastore/plugin_datastore_sqlite.c (working copy)
> @@ -29,6 +29,12 @@
> #include "plugin_datastore.h"
> #include <sqlite3.h>
> +/* For compatibility with older versions of sqlite3 */
> +#if !defined(HAVE_SQLITE3_INT64) && defined(HAVE_SQLITE_INT64)
> +typedef sqlite_int64 sqlite3_int64;
> +typedef sqlite_uint64 sqlite3_uint64;
> #define DEBUG_SQLITE GNUNET_NO
> Index: configure.ac
> --- configure.ac (revision 12650)
> +++ configure.ac (working copy)
> @@ -312,6 +312,11 @@
> +if test x$sqlite = xtrue
> + AC_CHECK_TYPES([sqlite3_int64,sqlite_int64],,,[#include <sqlite3.h>])
> # test for postgres
> AC_MSG_CHECKING(for postgres)
> Bug-GNUnet mailing list
> Bug-GNUnet@... > http://lists.gnu.org/mailman/listinfo/bug-gnunet