1 files changed, 0 insertions, 42 deletions
diff --git a/mail/bogofilter/files/patch-src_datastore__tdb.c b/mail/bogofilter/files/patch-src_datastore__tdb.c
deleted file mode 100644
index 549b1de7b689..000000000000
--- a/mail/bogofilter/files/patch-src_datastore__tdb.c
+++ /dev/null
@@ -1,42 +0,0 @@
---- a/src/datastore_tdb.c	31 May 2005 15:04:05 -0000	1.51.22.1
-+++ b/src/datastore_tdb.c	10 Jun 2005 08:19:30 -0000	1.51.22.2
-@@ -268,20 +268,34 @@
-     dbv_t dbv_key, dbv_data;
-     userdata_t *hookdata = userdata;
- 
--    /* Question: Is there a way to avoid using malloc/free? */
--
--    /* switch to "dbv_t *" variables */
-+    /* TDB happily returns data from odd addresses, so we need to
-+     * memcpy() everything to properly aligned storage (malloc() is
-+     * fine) if our callee wishes to perform word-sized access - without
-+     * memcpy(), the callee will die with SIGBUS (SPARC, m68k) or get
-+     * extremely slow (i386). */
-+
-+    /* XXX FIXME: Possible optimization if this function is only used by
-+     * one caller at a time (i. e. no threads): allocate buffers
-+     * statically and reuse them as long as they are of sufficient size
-+     * and reallocate otherwise. */
-+
-+    /* copy key */
-+    /* XXX FIXME: do we really need to use C-string compatible keys?
-+     * Looks wasteful. */
-     dbv_key.leng = key.dsize;
-     dbv_key.data = xmalloc(dbv_key.leng+1);
-     memcpy(dbv_key.data, key.dptr, dbv_key.leng);
-     ((char *)dbv_key.data)[dbv_key.leng] = '\0';
- 
--    dbv_data.data = data.dptr;
--    dbv_data.leng = data.dsize;		/* read count */
-+    /* copy data */
-+    dbv_data.leng = data.dsize;
-+    dbv_data.data = xmalloc(dbv_data.leng);
-+    memcpy(dbv_data.data, data.dptr, dbv_data.leng);
- 
-     /* call user function */
-     rc = hookdata->hook(&dbv_key, &dbv_data, hookdata->userdata);
- 
-+    xfree(dbv_data.data);
-     xfree(dbv_key.data);
- 
-     return rc;