Changeset: 7a616de6829a for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=7a616de6829a
Modified Files:
gdk/ChangeLog
gdk/gdk.h
gdk/gdk_bat.c
gdk/gdk_batop.mx
gdk/gdk_calc.c
gdk/gdk_relop.mx
gdk/gdk_scanselect.mx
monetdb5/extras/rdf/rdf_shredder.mx
monetdb5/modules/kernel/aggr_be_minmax.mx
monetdb5/modules/kernel/aggr_bge_minmax.mx
monetdb5/modules/kernel/algebra.mx
monetdb5/modules/mal/batExtensions.c
sql/backends/monet5/vaults/mseed.c
sql/storage/bat/bat_utils.c
Branch: default
Log Message:
Disallow BAT-of-BATs.
diffs (truncated from 1091 to 300 lines):
diff --git a/gdk/ChangeLog b/gdk/ChangeLog
--- a/gdk/ChangeLog
+++ b/gdk/ChangeLog
@@ -1,3 +1,8 @@
# ChangeLog file for MonetDB
# This file is updated with Maddlog
+* Tue Jul 17 2012 Sjoerd Mullender <sjo...@acm.org>
+- BAT-of-BATs is no longer allowed. It was already not allowed to
+ make these types of BATs persistent, but now they can't be created at
+ all anymore.
+
diff --git a/gdk/gdk.h b/gdk/gdk.h
--- a/gdk/gdk.h
+++ b/gdk/gdk.h
@@ -2133,12 +2133,6 @@ gdk_export oid OIDnew(oid inc);
* @item BAT*
* @tab
* BAThash (BAT *b, BUN masksize)
- * @item BAT *
- * @tab
- * BAThashsplit (BAT *b, BUN n, int unary)
- * @item BAT *
- * @tab
- * BATrangesplit (BAT *b, int n)
* @end multitable
*
* The current BAT implementation supports one search accelerator:
@@ -2147,14 +2141,8 @@ gdk_export oid OIDnew(oid inc);
* failure to create the supportive structures.
*
* The hash data structures are currently maintained during update operations.
- *
- * A BAT can be redistributed over n buckets using a hash
- * function with BAThashsplit. The return value is a list of BAT
- * pointers. Similarly, a range partitioning based is supported.
*/
gdk_export BAT *BAThash(BAT *b, BUN masksize);
-gdk_export BAT *BAThashsplit(BAT *b, BUN n, int unary);
-gdk_export BAT *BATrangesplit(BAT *b, BUN n, int unary);
gdk_export BAT *BAThashjoin(BAT *l, BAT *r, BUN estimate);
/* low level functions */
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -299,6 +299,8 @@ BATnew(int ht, int tt, BUN cap)
BATstore *bs;
assert(cap <= BUN_MAX);
+ assert(ht != TYPE_bat);
+ assert(tt != TYPE_bat);
ERRORcheck((ht < 0) || (ht > GDKatomcnt), "BATnew:ht error\n");
ERRORcheck((tt < 0) || (tt > GDKatomcnt), "BATnew:tt error\n");
@@ -763,6 +765,8 @@ BATcopy(BAT *b, int ht, int tt, int writ
BAT *bn = NULL;
BATcheck(b, "BATcopy");
+ assert(ht != TYPE_bat);
+ assert(tt != TYPE_bat);
cnt = b->batCount;
/* maybe a bit ugly to change the requested bat types?? */
@@ -2730,7 +2734,6 @@ BATgetaccess(BAT *b)
#define check_type(tp) \
do { \
if (ATOMisdescendant((tp), TYPE_ptr) || \
- ATOMisdescendant((tp), TYPE_bat) || \
BATatoms[tp].atomUnfix || \
BATatoms[tp].atomFix) { \
GDKerror("BATmode: %s type implies that %s[%s,%s] " \
@@ -2817,6 +2820,7 @@ BATassertHeadProps(BAT *b)
assert(b != NULL);
assert(b->htype >= TYPE_void);
assert(b->htype < GDKatomcnt);
+ assert(b->htype != TYPE_bat);
cmpf = BATatoms[b->htype].atomCmp;
nilp = ATOMnilptr(b->htype);
diff --git a/gdk/gdk_batop.mx b/gdk/gdk_batop.mx
--- a/gdk/gdk_batop.mx
+++ b/gdk/gdk_batop.mx
@@ -763,9 +763,8 @@ BATreplace(BAT *b, BAT *n, bit force)
* void-columns (in this case, the seqbase has to be recomputed in the result).
*
* Note that the BATslice() is used indirectly as well as a special
- * case for BATselect (range selection on sorted column), BATrangesplit
- * (fragmentation on sorted column) and BATsemijoin (when two dense columns
- * are semijoined).
+ * case for BATselect (range selection on sorted column) and
+ * BATsemijoin (when two dense columns are semijoined).
*
* NOTE new semantics, the selected range is excluding the high value.
*/
@@ -2002,320 +2001,6 @@ BATrevert(BAT *b)
}
/*
- * @+ BAT partitioning
- * For distributed processing we support hash and range
- * partitioning operators: BATsplithash and BATsplitrange.
- *
- * The part_bat function creates a partition BAT.
- */
-static BAT *
-part_bat(BAT *b, int ht, int tt, BUN expected_size, int respect_order)
-{
- BAT *bn = BATnew(ht, tt, (BUN) ((double) expected_size * BATMARGIN));
-
- if (bn) {
- BATkey(bn, BAThkey(b));
- BATkey(BATmirror(bn), BATtkey(b));
- bn->hsorted = respect_order && BAThordered(b);
- bn->tsorted = respect_order && BATtordered(b);
- bn->hrevsorted = respect_order && BAThrevordered(b);
- bn->trevsorted = respect_order && BATtrevordered(b);
- bn->H->nonil = b->H->nonil;
- bn->T->nonil = b->T->nonil;
- }
- return bn;
-}
-
-/*
- * @- hash partitioning
- */
-#define BUNhash(bx,hx,tx)\
- /* assert(n <= 0x40000000); */\
- i = (int)(HASHprobe(&h,tx)%n); \
- if ((r = BUNfnd(bx, &i)) != BUN_NONE){\
- bat bid = *(bat*)BUNtloc(metabati,r); \
- bunfastins(BBPdescriptor(bid), hx, tx);\
- }
-
-BAT *
-BAThashsplit(BAT *b, BUN n, int unary)
-{
- BAT *metabat, *bn, *bf;
- BUN r;
- Hash h;
- BUN cnt;
- int i = 0;
-
- /* assert(n <= 0x40000000); */
- BATcheck(b, "BAThashsplit");
- if (n > BATcount(b)) {
- PROPDEBUG THRprintf(GDKout, "#BAThashsplit: reduced number of
ranges (" BUNFMT ") to number of tuples (" BUNFMT ").\n", n, BATcount(b));
- n = BATcount(b);
- }
- if (n < 1) {
- GDKerror("BAThashsplit: number of ranges must not be less than
1!\n");
- return 0;
- }
-
- metabat = BATnew(TYPE_int, TYPE_bat, n);
- if (metabat == NULL)
- return NULL;
- bn = unary ? VIEWhead_(b, b->batRestricted) : b;
- if (n <= 1) {
- if (BUNins(metabat, &i, &bn->batCacheid, FALSE) == NULL)
- goto bunins_failed;
- } else {
- BUN p, q;
- BATiter metabati;
-
- for (i = 2; i < (int) n; i *= 2)
- ;
- h.mask = i - 1;
- h.type = BATttype(b);
- cnt = (BUN) (BATMARGIN * (double) BATbuncount(b) / (double) n);
- for (i = 0; i < (int) n; i++) {
- bf = part_bat(bn, BAThtype(bn), BATttype(bn), cnt,
TRUE);
- if (bf == NULL) {
- BBPreclaim(metabat);
- return NULL;
- }
- if (BUNins(metabat, &i, &bf->batCacheid, FALSE) == NULL)
- goto bunins_failed;
- }
- metabati = bat_iterator(metabat);
- updateloop(metabat, b, BUNhash);
- BATloop(metabat, p, q) {
- bat bt = *(bat *) BUNtail(metabati, p);
-
- BBPunfix(bt);
- }
- }
- return metabat;
- bunins_failed:
- BBPreclaim(metabat);
- return NULL;
-}
-
-/*
- * Range partitioning ensures that identical values appear in one
- * partition only. The routine also tries to deliver partitions of
- * uniform size.
- */
-BAT *
-BATrangesplit(BAT *b, BUN n, int unary)
-{
- BAT *metabat, *slice, *histo, *bf = NULL, *bn, *m;
- int target, tpe, *sizes = NULL;
- int zz = 0;
- BUN yy = 0;
- ptr *seps = NULL, nilval;
- BUN r, s;
- BATiter histoi, bi = bat_iterator(b);
- dbl scale;
- BUN thorough = (n <= 1 || BATtvoid(b) || BATtordered(b)) ? 1 : 10;
-
- BATcheck(b, "BATrangesplit");
- if (n > BATcount(b)) {
- PROPDEBUG THRprintf(GDKout, "#BATrangesplit: reduced number of
ranges (" BUNFMT ") to number of tuples (" BUNFMT ").\n", n, BATcount(b));
- n = BATcount(b);
- }
- if (n < 1) {
- GDKerror("BAThashsplit: number of ranges must not be less than
1!\n");
- return 0;
- }
-
- /* assert(BATcount(b)/n <= 0x7fffffff); */
- bn = unary ? VIEWhead_(b, b->batRestricted) : b;
- m = BATmirror(b);
- metabat = BATnew(BATttype(b), TYPE_bat, n);
- BATcheck(metabat, "BATrangesplit 2");
- nilval = ATOMnilptr(BATttype(b));
-
- /*
- * We use sampling to determine bucket sizes.
- * Uniform bucket sizes are the ideal to be achieved.
- * If necessary though, we deliver less than n buckets.
- */
- slice = BATsample(b, MIN(MAX(30 * n * thorough, 100 * thorough),
BATcount(b)));
- histo = BAThistogram(slice);
- target = (int) (BATcount(b) / n); /* see assert above */
- scale = ((dbl) BATcount(b)) / ((dbl) BATcount(slice));
- BBPreclaim(slice);
- if ((sizes = (int *) GDKmalloc(2 * n * sizeof(int))) == NULL)
- goto bunins_failed;
- if ((seps = (ptr *) GDKmalloc(2 * n * sizeof(ptr))) == NULL)
- goto bunins_failed;
- /*
- * Use the histogram to determine good split boundaries on b.
- */
- BATorder(histo);
- histoi = bat_iterator(histo);
- BATloop(histo, r, s) {
- int cnt = *(int *) BUNtloc(histoi, r);
- int add = (int) (scale * cnt);
-
- if (zz + add > target) {
- if ((zz + add - target) < (target - zz)) {
- sizes[yy] = zz + add;
- seps[yy] = ATOMdup(histo->htype,
BUNhead(histoi, r));
- add = 0;
- } else {
- sizes[yy] = zz;
- seps[yy] = ATOMdup(histo->htype,
BUNhead(histoi, (r - 1)));
- }
- zz = 0;
- yy++;
- }
- zz += add;
- }
- if (yy) {
- if ((sizes[yy - 1] + zz - target) > (target - zz)) {
- sizes[yy] = zz;
- } else {
- yy--; /* join with the last */
- }
- }
- seps[yy] = nilval;
- BBPreclaim(histo);
-
- if (n > 1 && n != yy + 1) {
- PROPDEBUG THRprintf(GDKout, "#rangesplit: delivering " BUNFMT "
instead of " BUNFMT " fragments\n", yy + 1, n);
- n = yy + 1;
- }
- /*
- * CASE 1: just one bucket.
- * This is done without copying b.
- */
- if (n <= 1) {
- if (BUNins(metabat, nilval, &bn->batCacheid, FALSE) == NULL)
- goto bunins_failed;
-
- /*
- * CASE 2: sorted on fragmentation column.
- * We can again avoid copying, by giving slices (views) on the source
BAT.
- * Virtual oids (void) is a special subcase with positional lookup
instead
- * of binary search.
- */
- } else if (BATtvoid(b) || BATtordered(b)) {
- BUN l, h = 0, o = BUNfirst(b);
-
- for (yy = 0; yy < n; yy++) {
- l = h;
- if (yy == n - 1) {
- r = BUNlast(m);
_______________________________________________
Checkin-list mailing list
Checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
