+ * Revision 1.22 1999-02-18 12:49:34 adam
+ * Changed file naming scheme for register files as well as record
+ * store/index files.
+ *
+ * Revision 1.21 1999/02/02 14:51:03 adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.20 1998/01/12 15:04:08 adam
+ * The test option (-s) only uses read-lock (and not write lock).
+ *
+ * Revision 1.19 1997/09/17 12:19:16 adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.18 1997/07/15 16:28:42 adam
+ * Bug fix: storeData didn't work with files with multiple records.
+ * Bug fix: fixed memory management with records; not really well
+ * thought through.
+ *
+ * Revision 1.17 1997/02/12 20:39:46 adam
+ * Implemented options -f <n> that limits the log to the first <n>
+ * records.
+ * Changed some log messages also.
+ *
+ * Revision 1.16 1996/06/04 10:19:00 adam
+ * Minor changes - removed include of ctype.h.
+ *
+ * Revision 1.15 1996/05/13 14:23:06 adam
+ * Work on compaction of set/use bytes in dictionary.
+ *
+ * Revision 1.14 1996/02/01 20:48:15 adam
+ * The total size of records are always checked in rec_cache_insert to
+ * reduce memory usage.
+ *
+ * Revision 1.13 1995/12/11 09:12:49 adam
+ * The rec_get function returns NULL if record doesn't exist - will
+ * happen in the server if the result set records have been deleted since
+ * the creation of the set (i.e. the search).
+ * The server saves a result temporarily if it is 'volatile', i.e. the
+ * set is register dependent.
+ *
+ * Revision 1.12 1995/12/07 17:38:47 adam
+ * Work locking mechanisms for concurrent updates/commit.
+ *
+ * Revision 1.11 1995/12/06 13:58:26 adam
+ * Improved flushing of records - all flushes except the last one
+ * don't write the last accessed. Also flush takes place if record
+ * info occupy more than about 256k.
+ *
+ * Revision 1.10 1995/12/06 12:41:24 adam