Skip to content

Instantly share code, notes, and snippets.

@mrzzzrm

mrzzzrm/gist:82ac2721e87eb5a6645a5482c0ab7f7c

Created April 16, 2019 15:18
Show Gist options
  • Save mrzzzrm/82ac2721e87eb5a6645a5482c0ab7f7c to your computer and use it in GitHub Desktop.
Save mrzzzrm/82ac2721e87eb5a6645a5482c0ab7f7c to your computer and use it in GitHub Desktop.
Download ZIP
Raw
gistfile1.txt
Only in ../tpch-dbgen/: answers
Only in ../tpch-dbgen/: bcd2.c
Only in ../tpch-dbgen/: bcd2.h
Only in ../tpch-dbgen/: bcd2.o
diff -p third_party/tpch-dbgen/bm_utils.c ../tpch-dbgen/bm_utils.c
*** third_party/tpch-dbgen/bm_utils.c 2019-04-16 17:06:53.528759641 +0200
--- ../tpch-dbgen/bm_utils.c 2019-04-16 11:39:42.886691071 +0200
***************
*** 67,73 ****
#endif /* HP */
#include <ctype.h>
#include <math.h>
! #if !defined(_POSIX_SOURCE) && !defined(__APPLE__)
#include <malloc.h>
#endif /* POSIX_SOURCE */
#include <fcntl.h>
--- 67,73 ----
#endif /* HP */
#include <ctype.h>
#include <math.h>
! #ifndef _POSIX_SOURCE
#include <malloc.h>
#endif /* POSIX_SOURCE */
#include <fcntl.h>
*************** pick_str(distribution *s, int c, char *t
*** 230,238 ****
RANDOM(j, 1, s->list[s->count - 1].weight, c);
while (s->list[i].weight < j)
i++;
! // HYRISE: Avoid redundantly obtaining the length of a string
! if (!s->list[i].len) s->list[i].len = strlen(s->list[i].text);
! memcpy(target, s->list[i].text, s->list[i].len + 1);
return(i);
}
--- 230,236 ----
RANDOM(j, 1, s->list[s->count - 1].weight, c);
while (s->list[i].weight < j)
i++;
! strcpy(target, s->list[i].text);
return(i);
}
*************** long weight,
*** 372,378 ****
}
target->permute = (long *)NULL;
fclose(fp);
-
return;
}
--- 370,375 ----
Only in ../tpch-dbgen/: bm_utils.o
Only in ../tpch-dbgen/: BUGS
diff -p third_party/tpch-dbgen/build.c ../tpch-dbgen/build.c
*** third_party/tpch-dbgen/build.c 2019-04-16 17:08:35.315673948 +0200
--- ../tpch-dbgen/build.c 2019-04-16 11:39:42.886691071 +0200
*************** extern adhoc_t adhocs[];
*** 65,73 ****
#define TEXT(avg, sd, tgt) dbg_text(tgt, (int)(avg * V_STR_LOW),(int)(avg * V_STR_HGH), sd)
static void gen_phone PROTO((DSS_HUGE ind, char *target, long seed));
- // HYRISE: Made public so we can free the allocated memory later
- char **asc_date = NULL;
-
DSS_HUGE
rpb_routine(DSS_HUGE p)
{
--- 65,70 ----
*************** mk_order(DSS_HUGE index, order_t * o, lo
*** 160,165 ****
--- 157,163 ----
DSS_HUGE c_date;
DSS_HUGE clk_num;
DSS_HUGE supp_num;
+ static char **asc_date = NULL;
char tmp_str[2];
char **mk_ascdate PROTO((void));
int delta = 1;
*************** mk_order(DSS_HUGE index, order_t * o, lo
*** 188,195 ****
RANDOM(tmp_date, O_ODATE_MIN, O_ODATE_MAX, O_ODATE_SD);
! // HYRISE: We know that dates are always yyyy-mm-dd + \0.
! memcpy(o->odate, asc_date[tmp_date - STARTDATE], 11);
pick_str(&o_priority_set, O_PRIO_SD, o->opriority);
RANDOM(clk_num, 1, MAX((scale * O_CLRK_SCL), O_CLRK_SCL), O_CLRK_SD);
--- 186,192 ----
RANDOM(tmp_date, O_ODATE_MIN, O_ODATE_MAX, O_ODATE_SD);
! strcpy(o->odate, asc_date[tmp_date - STARTDATE]);
pick_str(&o_priority_set, O_PRIO_SD, o->opriority);
RANDOM(clk_num, 1, MAX((scale * O_CLRK_SCL), O_CLRK_SCL), O_CLRK_SD);
*************** mk_order(DSS_HUGE index, order_t * o, lo
*** 240,249 ****
RANDOM(r_date, L_RDTE_MIN, L_RDTE_MAX, L_RDTE_SD);
r_date += s_date;
! // HYRISE: We know that dates are always yyyy-mm-dd + \0.
! memcpy(o->l[lcnt].sdate, asc_date[s_date - STARTDATE], 11);
! memcpy(o->l[lcnt].cdate, asc_date[c_date - STARTDATE], 11);
! memcpy(o->l[lcnt].rdate, asc_date[r_date - STARTDATE], 11);
if (julian(r_date) <= CURRENTDATE)
--- 237,246 ----
RANDOM(r_date, L_RDTE_MIN, L_RDTE_MAX, L_RDTE_SD);
r_date += s_date;
!
! strcpy(o->l[lcnt].sdate, asc_date[s_date - STARTDATE]);
! strcpy(o->l[lcnt].cdate, asc_date[c_date - STARTDATE]);
! strcpy(o->l[lcnt].rdate, asc_date[r_date - STARTDATE]);
if (julian(r_date) <= CURRENTDATE)
Only in ../tpch-dbgen/: build.o
Only in ../tpch-dbgen/: check_answers
Only in third_party/tpch-dbgen: CMakeLists.txt
Only in ../tpch-dbgen/: column_split.sh
Only in ../tpch-dbgen/: customer.tbl
Only in ../tpch-dbgen/: dbgen
Only in ../tpch-dbgen/: dbgen.dsp
Only in ../tpch-dbgen/: dists.dss
diff -p third_party/tpch-dbgen/driver.c ../tpch-dbgen/driver.c
*** third_party/tpch-dbgen/driver.c 2019-04-16 17:15:09.968328697 +0200
--- ../tpch-dbgen/driver.c 2019-04-16 11:53:30.786520451 +0200
***************
*** 102,108 ****
#include "dss.h"
#include "dsstypes.h"
! #include "tpch_dbgen.h"
extern int optind, opterr;
--- 102,118 ----
#include "dss.h"
#include "dsstypes.h"
!
! /*
! * Function prototypes
! */
! void usage (void);
! void kill_load (void);
! int pload (int tbl);
! void gen_tbl (int tnum, DSS_HUGE start, DSS_HUGE count, long upd_num);
! int pr_drange (int tbl, DSS_HUGE min, DSS_HUGE cnt, long num);
! int set_files (int t, int pload);
! int partial (int, int);
extern int optind, opterr;
*************** static int bTableSet = 0;
*** 141,146 ****
--- 151,170 ----
*/
/*
+ * flat file print functions; used with -F(lat) option
+ */
+ int pr_cust (customer_t * c, int mode);
+ int pr_line (order_t * o, int mode);
+ int pr_order (order_t * o, int mode);
+ int pr_part (part_t * p, int mode);
+ int pr_psupp (part_t * p, int mode);
+ int pr_supp (supplier_t * s, int mode);
+ int pr_order_line (order_t * o, int mode);
+ int pr_part_psupp (part_t * p, int mode);
+ int pr_nation (code_t * c, int mode);
+ int pr_region (code_t * c, int mode);
+
+ /*
* seed generation functions; used with '-O s' option
*/
long sd_cust (int child, DSS_HUGE skip_count);
*************** long sd_supp (int child, DSS_HUGE skip_c
*** 152,180 ****
long sd_order_line (int child, DSS_HUGE skip_count);
long sd_part_psupp (int child, DSS_HUGE skip_count);
- // HYRISE: removed loader-functions (pr_part, ...) as we do not need them. We don't print
- // the tables to text.
tdef tdefs[] =
{
{"part.tbl", "part table", 200000,
! NULL, sd_part, PSUPP, 0},
{"partsupp.tbl", "partsupplier table", 200000,
! NULL, sd_psupp, NONE, 0},
{"supplier.tbl", "suppliers table", 10000,
! NULL, sd_supp, NONE, 0},
{"customer.tbl", "customers table", 150000,
! NULL, sd_cust, NONE, 0},
{"orders.tbl", "order table", 150000,
! NULL, sd_order, LINE, 0},
{"lineitem.tbl", "lineitem table", 150000,
! NULL, sd_line, NONE, 0},
{"orders.tbl", "orders/lineitem tables", 150000,
! NULL, sd_order, LINE, 0},
{"part.tbl", "part/partsupplier tables", 200000,
! NULL, sd_part, PSUPP, 0},
! {"nation.tbl", "nation table", NATIONS_COUNT,
! NULL, NO_LFUNC, NONE, 0},
! {"region.tbl", "region table", REGIONS_COUNT,
! NULL, NO_LFUNC, NONE, 0},
};
--- 176,800 ----
long sd_order_line (int child, DSS_HUGE skip_count);
long sd_part_psupp (int child, DSS_HUGE skip_count);
tdef tdefs[] =
{
{"part.tbl", "part table", 200000,
! pr_part, sd_part, PSUPP, 0},
{"partsupp.tbl", "partsupplier table", 200000,
! pr_psupp, sd_psupp, NONE, 0},
{"supplier.tbl", "suppliers table", 10000,
! pr_supp, sd_supp, NONE, 0},
{"customer.tbl", "customers table", 150000,
! pr_cust, sd_cust, NONE, 0},
{"orders.tbl", "order table", 150000,
! pr_order, sd_order, LINE, 0},
{"lineitem.tbl", "lineitem table", 150000,
! pr_line, sd_line, NONE, 0},
{"orders.tbl", "orders/lineitem tables", 150000,
! pr_order_line, sd_order, LINE, 0},
{"part.tbl", "part/partsupplier tables", 200000,
! pr_part_psupp, sd_part, PSUPP, 0},
! {"nation.tbl", "nation table", NATIONS_MAX,
! pr_nation, NO_LFUNC, NONE, 0},
! {"region.tbl", "region table", NATIONS_MAX,
! pr_region, NO_LFUNC, NONE, 0},
};
+ /*
+ * re-set default output file names
+ */
+ int
+ set_files (int i, int pload)
+ {
+ char line[80], *new_name;
+
+ if (table & (1 << i))
+ child_table:
+ {
+ if (pload != -1)
+ sprintf (line, "%s.%d", tdefs[i].name, pload);
+ else
+ {
+ printf ("Enter new destination for %s data: ",
+ tdefs[i].name);
+ if (fgets (line, sizeof (line), stdin) == NULL)
+ return (-1);;
+ if ((new_name = strchr (line, '\n')) != NULL)
+ *new_name = '\0';
+ if ((int)strlen (line) == 0)
+ return (0);
+ }
+ new_name = (char *) malloc ((int)strlen (line) + 1);
+ MALLOC_CHECK (new_name);
+ strcpy (new_name, line);
+ tdefs[i].name = new_name;
+ if (tdefs[i].child != NONE)
+ {
+ i = tdefs[i].child;
+ tdefs[i].child = NONE;
+ goto child_table;
+ }
+ }
+
+ return (0);
+ }
+
+
+
+ /*
+ * read the distributions needed in the benchamrk
+ */
+ void
+ load_dists (void)
+ {
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "p_cntr", &p_cntr_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "colors", &colors);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "p_types", &p_types_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "nations", &nations);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "regions", &regions);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "o_oprio",
+ &o_priority_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "instruct",
+ &l_instruct_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "smode", &l_smode_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "category",
+ &l_category_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "rflag", &l_rflag_set);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "msegmnt", &c_mseg_set);
+
+ /* load the distributions that contain text generation */
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "nouns", &nouns);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "verbs", &verbs);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "adjectives", &adjectives);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "adverbs", &adverbs);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "auxillaries", &auxillaries);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "terminators", &terminators);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "articles", &articles);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "prepositions", &prepositions);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "grammar", &grammar);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "np", &np);
+ read_dist (env_config (DIST_TAG, DIST_DFLT), "vp", &vp);
+
+ }
+
+ /*
+ * generate a particular table
+ */
+ void
+ gen_tbl (int tnum, DSS_HUGE start, DSS_HUGE count, long upd_num)
+ {
+ static order_t o;
+ supplier_t supp;
+ customer_t cust;
+ part_t part;
+ code_t code;
+ static int completed = 0;
+ DSS_HUGE i;
+
+ DSS_HUGE rows_per_segment=0;
+ DSS_HUGE rows_this_segment=-1;
+ DSS_HUGE residual_rows=0;
+
+ if (insert_segments)
+ {
+ rows_per_segment = count / insert_segments;
+ residual_rows = count - (rows_per_segment * insert_segments);
+ }
+
+ for (i = start; count; count--, i++)
+ {
+ LIFENOISE (1000, i);
+ row_start(tnum);
+
+ switch (tnum)
+ {
+ case LINE:
+ case ORDER:
+ case ORDER_LINE:
+ mk_order (i, &o, upd_num % 10000);
+
+ if (insert_segments && (upd_num > 0))
+ if((upd_num / 10000) < residual_rows)
+ {
+ if((++rows_this_segment) > rows_per_segment)
+ {
+ rows_this_segment=0;
+ upd_num += 10000;
+ }
+ }
+ else
+ {
+ if((++rows_this_segment) >= rows_per_segment)
+ {
+ rows_this_segment=0;
+ upd_num += 10000;
+ }
+ }
+
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&o, upd_num);
+ break;
+ case SUPP:
+ mk_supp (i, &supp);
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&supp, upd_num);
+ break;
+ case CUST:
+ mk_cust (i, &cust);
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&cust, upd_num);
+ break;
+ case PSUPP:
+ case PART:
+ case PART_PSUPP:
+ mk_part (i, &part);
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&part, upd_num);
+ break;
+ case NATION:
+ mk_nation (i, &code);
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&code, 0);
+ break;
+ case REGION:
+ mk_region (i, &code);
+ if (set_seeds == 0)
+ tdefs[tnum].loader(&code, 0);
+ break;
+ }
+ row_stop(tnum);
+ if (set_seeds && (i % tdefs[tnum].base) < 2)
+ {
+ printf("\nSeeds for %s at rowcount %ld\n", tdefs[tnum].comment, i);
+ dump_seeds(tnum);
+ }
+ }
+ completed |= 1 << tnum;
+ }
+
+
+
+ void
+ usage (void)
+ {
+ fprintf (stderr, "%s\n%s\n\t%s\n%s %s\n\n",
+ "USAGE:",
+ "dbgen [-{vf}][-T {pcsoPSOL}]",
+ "[-s <scale>][-C <procs>][-S <step>]",
+ "dbgen [-v] [-O m] [-s <scale>]",
+ "[-U <updates>]");
+ fprintf (stderr, "Basic Options\n===========================\n");
+ fprintf (stderr, "-C <n> -- separate data set into <n> chunks (requires -S, default: 1)\n");
+ fprintf (stderr, "-f -- force. Overwrite existing files\n");
+ fprintf (stderr, "-h -- display this message\n");
+ fprintf (stderr, "-q -- enable QUIET mode\n");
+ fprintf (stderr, "-s <n> -- set Scale Factor (SF) to <n> (default: 1) \n");
+ fprintf (stderr, "-S <n> -- build the <n>th step of the data/update set (used with -C or -U)\n");
+ fprintf (stderr, "-U <n> -- generate <n> update sets\n");
+ fprintf (stderr, "-v -- enable VERBOSE mode\n");
+ fprintf (stderr, "\nAdvanced Options\n===========================\n");
+ fprintf (stderr, "-b <s> -- load distributions for <s> (default: dists.dss)\n");
+ fprintf (stderr, "-d <n> -- split deletes between <n> files (requires -U)\n");
+ fprintf (stderr, "-i <n> -- split inserts between <n> files (requires -U)\n");
+ fprintf (stderr, "-T c -- generate cutomers ONLY\n");
+ fprintf (stderr, "-T l -- generate nation/region ONLY\n");
+ fprintf (stderr, "-T L -- generate lineitem ONLY\n");
+ fprintf (stderr, "-T n -- generate nation ONLY\n");
+ fprintf (stderr, "-T o -- generate orders/lineitem ONLY\n");
+ fprintf (stderr, "-T O -- generate orders ONLY\n");
+ fprintf (stderr, "-T p -- generate parts/partsupp ONLY\n");
+ fprintf (stderr, "-T P -- generate parts ONLY\n");
+ fprintf (stderr, "-T r -- generate region ONLY\n");
+ fprintf (stderr, "-T s -- generate suppliers ONLY\n");
+ fprintf (stderr, "-T S -- generate partsupp ONLY\n");
+ fprintf (stderr,
+ "\nTo generate the SF=1 (1GB), validation database population, use:\n");
+ fprintf (stderr, "\tdbgen -vf -s 1\n");
+ fprintf (stderr, "\nTo generate updates for a SF=1 (1GB), use:\n");
+ fprintf (stderr, "\tdbgen -v -U 1 -s 1\n");
+ }
+
+ /*
+ * int partial(int tbl, int s) -- generate the s-th part of the named tables data
+ */
+ int
+ partial (int tbl, int s)
+ {
+ DSS_HUGE rowcnt;
+ DSS_HUGE extra;
+
+ if (verbose > 0)
+ {
+ fprintf (stderr, "\tStarting to load stage %d of %ld for %s...",
+ s, children, tdefs[tbl].comment);
+ }
+
+ set_files (tbl, s);
+
+ rowcnt = set_state(tbl, scale, children, s, &extra);
+
+ if (s == children)
+ gen_tbl (tbl, rowcnt * (s - 1) + 1, rowcnt + extra, upd_num);
+ else
+ gen_tbl (tbl, rowcnt * (s - 1) + 1, rowcnt, upd_num);
+
+ if (verbose > 0)
+ fprintf (stderr, "done.\n");
+
+ return (0);
+ }
+
+ void
+ process_options (int count, char **vector)
+ {
+ int option;
+ FILE *pF;
+
+ while ((option = getopt (count, vector,
+ "b:C:d:fi:hO:P:qs:S:T:U:v")) != -1)
+ switch (option)
+ {
+ case 'b': /* load distributions from named file */
+ d_path = (char *)malloc((int)strlen(optarg) + 1);
+ MALLOC_CHECK(d_path);
+ strcpy(d_path, optarg);
+ if ((pF = fopen(d_path, "r")) == NULL)
+ {
+ fprintf(stderr, "ERROR: Invalid argument to -b");
+ exit(-1);
+ }
+ else
+ fclose(pF);
+
+ break;
+ case 'C':
+ children = atoi (optarg);
+ break;
+ case 'd':
+ delete_segments = atoi (optarg);
+ break;
+ case 'f': /* blind overwrites; Force */
+ force = 1;
+ break;
+ case 'i':
+ insert_segments = atoi (optarg);
+ break;
+ case 'q': /* all prompts disabled */
+ verbose = -1;
+ break;
+ case 's': /* scale by Percentage of base rowcount */
+ case 'P': /* for backward compatibility */
+ flt_scale = atof (optarg);
+ if (flt_scale < MIN_SCALE)
+ {
+ int i;
+ int int_scale;
+
+ scale = 1;
+ int_scale = (int)(1000 * flt_scale);
+ for (i = PART; i < REGION; i++)
+ {
+ tdefs[i].base = (DSS_HUGE)(int_scale * tdefs[i].base)/1000;
+ if (tdefs[i].base < 1)
+ tdefs[i].base = 1;
+ }
+ }
+ else
+ scale = (long) flt_scale;
+ if (scale > MAX_SCALE)
+ {
+ fprintf (stderr, "%s %5.0f %s\n\t%s\n\n",
+ "NOTE: Data generation for scale factors >",
+ MAX_SCALE,
+ "GB is still in development,",
+ "and is not yet supported.\n");
+ fprintf (stderr,
+ "Your resulting data set MAY NOT BE COMPLIANT!\n");
+ }
+ break;
+ case 'S': /* generate a particular STEP */
+ step = atoi (optarg);
+ break;
+ case 'U': /* generate flat files for update stream */
+ updates = atoi (optarg);
+ break;
+ case 'v': /* life noises enabled */
+ verbose = 1;
+ break;
+ case 'T': /* generate a specifc table */
+ switch (*optarg)
+ {
+ case 'c': /* generate customer ONLY */
+ table = 1 << CUST;
+ bTableSet = 1;
+ break;
+ case 'L': /* generate lineitems ONLY */
+ table = 1 << LINE;
+ bTableSet = 1;
+ break;
+ case 'l': /* generate code table ONLY */
+ table = 1 << NATION;
+ table |= 1 << REGION;
+ bTableSet = 1;
+ break;
+ case 'n': /* generate nation table ONLY */
+ table = 1 << NATION;
+ bTableSet = 1;
+ break;
+ case 'O': /* generate orders ONLY */
+ table = 1 << ORDER;
+ bTableSet = 1;
+ break;
+ case 'o': /* generate orders/lineitems ONLY */
+ table = 1 << ORDER_LINE;
+ bTableSet = 1;
+ break;
+ case 'P': /* generate part ONLY */
+ table = 1 << PART;
+ bTableSet = 1;
+ break;
+ case 'p': /* generate part/partsupp ONLY */
+ table = 1 << PART_PSUPP;
+ bTableSet = 1;
+ break;
+ case 'r': /* generate region table ONLY */
+ table = 1 << REGION;
+ bTableSet = 1;
+ break;
+ case 'S': /* generate partsupp ONLY */
+ table = 1 << PSUPP;
+ bTableSet = 1;
+ break;
+ case 's': /* generate suppliers ONLY */
+ table = 1 << SUPP;
+ bTableSet = 1;
+ break;
+ default:
+ fprintf (stderr, "Unknown table name %s\n",
+ optarg);
+ usage ();
+ exit (1);
+ }
+ break;
+ case 'O': /* optional actions */
+ switch (tolower (*optarg))
+ {
+ case 's': /* calibrate the RNG usage */
+ set_seeds = 1;
+ break;
+ default:
+ fprintf (stderr, "Unknown option name %s\n",
+ optarg);
+ usage ();
+ exit (1);
+ }
+ break;
+ default:
+ printf ("ERROR: option '%c' unknown.\n",
+ *(vector[optind] + 1));
+ case 'h': /* something unexpected */
+ fprintf (stderr,
+ "%s Population Generator (Version %d.%d.%d build %d)\n",
+ NAME, VERSION, RELEASE, PATCH, BUILD);
+ fprintf (stderr, "Copyright %s %s\n", TPC, C_DATES);
+ usage ();
+ exit (1);
+ }
+
+ return;
+ }
+
+ void validate_options(void)
+ {
+ // DBGenOptions, 3.1
+ if (children != 1)
+ {
+ if (updates != 0)
+ {
+ fprintf(stderr, "ERROR: -C is not valid when generating updates\n");
+ exit(-1);
+ }
+ if (step == -1)
+ {
+ fprintf(stderr, "ERROR: -S must be specified when generating data in multiple chunks\n");
+ exit(-1);
+ }
+ }
+
+ // DBGenOptions, 3.3
+ if (updates == 0)
+ {
+ if ((insert_segments != 0) || (delete_segments != 0))
+ {
+ fprintf(stderr, "ERROR: -d/-i are only valid when generating updates\n");
+ exit(-1);
+ }
+ }
+
+ // DBGenOptions, 3.9
+ if (step != -1)
+ {
+ if ((children == 1) && (updates == 0))
+ {
+ fprintf(stderr, "ERROR: -S is only valid when generating data in multiple chunks or generating updates\n");
+ exit(-1);
+ }
+ }
+
+ // DBGenOptions, 3.10
+ if (bTableSet && (updates != 0))
+ {
+ fprintf(stderr, "ERROR: -T not valid when generating updates\n");
+ exit(-1);
+ }
+
+ return;
+ }
+
+
+ /*
+ * MAIN
+ *
+ * assumes the existance of getopt() to clean up the command
+ * line handling
+ */
+ int
+ main (int ac, char **av)
+ {
+ DSS_HUGE i;
+
+ table = (1 << CUST) |
+ (1 << SUPP) |
+ (1 << NATION) |
+ (1 << REGION) |
+ (1 << PART_PSUPP) |
+ (1 << ORDER_LINE);
+ force = 0;
+ insert_segments=0;
+ delete_segments=0;
+ insert_orders_segment=0;
+ insert_lineitem_segment=0;
+ delete_segment=0;
+ verbose = 0;
+ set_seeds = 0;
+ scale = 1;
+ flt_scale = 1.0;
+ updates = 0;
+ step = -1;
+ tdefs[ORDER].base *=
+ ORDERS_PER_CUST; /* have to do this after init */
+ tdefs[LINE].base *=
+ ORDERS_PER_CUST; /* have to do this after init */
+ tdefs[ORDER_LINE].base *=
+ ORDERS_PER_CUST; /* have to do this after init */
+ children = 1;
+ d_path = NULL;
+
+ #ifdef NO_SUPPORT
+ signal (SIGINT, exit);
+ #endif /* NO_SUPPORT */
+ process_options (ac, av);
+ validate_options();
+ #if (defined(WIN32)&&!defined(_POSIX_))
+ for (i = 0; i < ac; i++)
+ {
+ spawn_args[i] = malloc (((int)strlen (av[i]) + 1) * sizeof (char));
+ MALLOC_CHECK (spawn_args[i]);
+ strcpy (spawn_args[i], av[i]);
+ }
+ spawn_args[ac] = NULL;
+ #endif
+
+ if (verbose >= 0)
+ {
+ fprintf (stderr,
+ "%s Population Generator (Version %d.%d.%d)\n",
+ NAME, VERSION, RELEASE, PATCH);
+ fprintf (stderr, "Copyright %s %s\n", TPC, C_DATES);
+ }
+
+ load_dists ();
+ #ifdef RNG_TEST
+ for (i=0; i <= MAX_STREAM; i++)
+ Seed[i].nCalls = 0;
+ #endif
+ /* have to do this after init */
+ tdefs[NATION].base = nations.count;
+ tdefs[REGION].base = regions.count;
+
+ /*
+ * updates are never parallelized
+ */
+ if (updates)
+ {
+ /*
+ * set RNG to start generating rows beyond SF=scale
+ */
+ set_state (ORDER, scale, 100, 101, &i);
+ rowcnt = (int)(tdefs[ORDER_LINE].base / 10000 * scale * UPD_PCT);
+ if (step > 0)
+ {
+ /*
+ * adjust RNG for any prior update generation
+ */
+ for (i=1; i < step; i++)
+ {
+ sd_order(0, rowcnt);
+ sd_line(0, rowcnt);
+ }
+ upd_num = step - 1;
+ }
+ else
+ upd_num = 0;
+
+ while (upd_num < updates)
+ {
+ if (verbose > 0)
+ fprintf (stderr,
+ "Generating update pair #%ld for %s",
+ upd_num + 1, tdefs[ORDER_LINE].comment);
+ insert_orders_segment=0;
+ insert_lineitem_segment=0;
+ delete_segment=0;
+ minrow = upd_num * rowcnt + 1;
+ gen_tbl (ORDER_LINE, minrow, rowcnt, upd_num + 1);
+ if (verbose > 0)
+ fprintf (stderr, "done.\n");
+ pr_drange (ORDER_LINE, minrow, rowcnt, upd_num + 1);
+ upd_num++;
+ }
+
+ exit (0);
+ }
+
+ /**
+ ** actual data generation section starts here
+ **/
+
+ /*
+ * traverse the tables, invoking the appropriate data generation routine for any to be built
+ */
+ for (i = PART; i <= REGION; i++)
+ if (table & (1 << i))
+ {
+ if (children > 1 && i < NATION)
+ {
+ partial ((int)i, step);
+ }
+ else
+ {
+ minrow = 1;
+ if (i < NATION)
+ rowcnt = tdefs[i].base * scale;
+ else
+ rowcnt = tdefs[i].base;
+ if (verbose > 0)
+ fprintf (stderr, "Generating data for %s %li", tdefs[i].comment, tdefs[i].base);
+ gen_tbl ((int)i, minrow, rowcnt, upd_num);
+ if (verbose > 0)
+ fprintf (stderr, "done.\n");
+ }
+ }
+
+ return (0);
+ }
Only in ../tpch-dbgen/: driver.o
Only in ../tpch-dbgen/: dss.ddl
diff -p third_party/tpch-dbgen/dss.h ../tpch-dbgen/dss.h
*** third_party/tpch-dbgen/dss.h 2019-04-16 17:06:26.193072689 +0200
--- ../tpch-dbgen/dss.h 2019-04-16 11:39:42.886691071 +0200
***************
*** 80,86 ****
#include "config.h"
#include "shared.h"
- #include "tpch_dbgen.h"
#include <stdio.h>
#include <stdlib.h>
--- 80,85 ----
***************
*** 112,117 ****
--- 111,119 ----
#define INTERNAL_ERROR(p) {fprintf(stderr,"%s", p);abort();}
#define LN_CNT 4
+ static char lnoise[4] = {'|', '/', '-', '\\' };
+ #define LIFENOISE(n, var) \
+ if (verbose > 0) fprintf(stderr, "%c\b", lnoise[(var%LN_CNT)])
#define MALLOC_CHECK(var) \
if ((var) == NULL) \
***************
*** 139,145 ****
#ifndef BITS_PER_LONG
#define BITS_PER_LONG 32
#define MAX_LONG 0x7FFFFFFF
- #define MIN_LONG (int32_t)0x80000000
#endif /* BITS_PER_LONG */
#define SPARSE_BITS 2
#define SPARSE_KEEP 3
--- 141,146 ----
***************
*** 148,154 ****
--- 149,170 ----
#define RANDOM(tgt, lower, upper, stream) dss_random(&tgt, lower, upper, stream)
#define RANDOM64(tgt, lower, upper, stream) dss_random64(&tgt, lower, upper, stream)
+
+
+ typedef struct
+ {
+ long weight;
+ char *text;
+ } set_member;
+
+ typedef struct
+ {
+ int count;
+ int max;
+ set_member *list;
+ long *permute;
+ } distribution;
/*
* some handy access functions
*/
*************** void dbg_text PROTO((char * t, int min,
*** 225,233 ****
--- 241,275 ----
#endif /* DECLARER */
+ EXTERN distribution nations;
+ EXTERN distribution nations2;
+ EXTERN distribution regions;
+ EXTERN distribution o_priority_set;
+ EXTERN distribution l_instruct_set;
+ EXTERN distribution l_smode_set;
+ EXTERN distribution l_category_set;
+ EXTERN distribution l_rflag_set;
+ EXTERN distribution c_mseg_set;
+ EXTERN distribution colors;
+ EXTERN distribution p_types_set;
+ EXTERN distribution p_cntr_set;
+
+ /* distributions that control text generation */
+ EXTERN distribution articles;
+ EXTERN distribution nouns;
+ EXTERN distribution adjectives;
+ EXTERN distribution adverbs;
+ EXTERN distribution prepositions;
+ EXTERN distribution verbs;
+ EXTERN distribution terminators;
+ EXTERN distribution auxillaries;
+ EXTERN distribution np;
+ EXTERN distribution vp;
+ EXTERN distribution grammar;
EXTERN long scale;
+ EXTERN int refresh;
EXTERN int resume;
EXTERN long verbose;
EXTERN long force;
Only in ../tpch-dbgen/: dss.ri
diff -p third_party/tpch-dbgen/dsstypes.h ../tpch-dbgen/dsstypes.h
*** third_party/tpch-dbgen/dsstypes.h 2019-04-16 09:37:49.006642496 +0200
--- ../tpch-dbgen/dsstypes.h 2019-04-16 11:39:42.886691071 +0200
*************** typedef struct
*** 128,133 ****
--- 128,134 ----
/* parts.c */
long mk_part PROTO((DSS_HUGE index, part_t * p));
+ int pr_part PROTO((part_t * part, int mode));
int ld_part PROTO((part_t * part, int mode));
typedef struct
Only in ../tpch-dbgen/: .git
Only in ../tpch-dbgen/: .gitignore
Only in ../tpch-dbgen/: HISTORY
Only in ../tpch-dbgen/: .idea
Only in third_party/tpch-dbgen: LICENSE
Only in ../tpch-dbgen/: lineitem.tbl
Only in ../tpch-dbgen/: load_stub.o
Only in ../tpch-dbgen/: makefile
Only in ../tpch-dbgen/: makefile.suite
Only in ../tpch-dbgen/: nation.tbl
Only in ../tpch-dbgen/: orders.tbl
Only in ../tpch-dbgen/: partsupp.tbl
Only in ../tpch-dbgen/: part.tbl
Only in ../tpch-dbgen/: permute.o
Only in ../tpch-dbgen/: PORTING.NOTES
Only in ../tpch-dbgen/: print.c
Only in ../tpch-dbgen/: print.o
Only in ../tpch-dbgen/: qgen
Only in ../tpch-dbgen/: qgen.c
Only in ../tpch-dbgen/: qgen.o
Only in ../tpch-dbgen/: qgen.vcproj
Only in ../tpch-dbgen/: queries
Only in ../tpch-dbgen/: reference
Only in ../tpch-dbgen/: region.tbl
diff -p third_party/tpch-dbgen/rnd.c ../tpch-dbgen/rnd.c
*** third_party/tpch-dbgen/rnd.c 2019-04-16 17:15:36.872136121 +0200
--- ../tpch-dbgen/rnd.c 2019-04-16 11:39:42.890690901 +0200
*************** row_stop(int t) \
*** 125,139 ****
for (i=0; i <= MAX_STREAM; i++)
if ((Seed[i].table == t) || (Seed[i].table == tdefs[t].child))
! {
! // HYRISE: commented out because set_seeds is considered false for us.
! // if (set_seeds && (Seed[i].usage > Seed[i].boundary))
! // {
! // fprintf(stderr, "\nSEED CHANGE: seed[%d].usage = %ld\n",
! // i, Seed[i].usage);
! // Seed[i].boundary = Seed[i].usage;
! // }
! // else
{
NthElement((Seed[i].boundary - Seed[i].usage), &Seed[i].value);
#ifdef RNG_TEST
--- 125,138 ----
for (i=0; i <= MAX_STREAM; i++)
if ((Seed[i].table == t) || (Seed[i].table == tdefs[t].child))
! {
! if (set_seeds && (Seed[i].usage > Seed[i].boundary))
! {
! fprintf(stderr, "\nSEED CHANGE: seed[%d].usage = %ld\n",
! i, Seed[i].usage);
! Seed[i].boundary = Seed[i].usage;
! }
! else
{
NthElement((Seed[i].boundary - Seed[i].usage), &Seed[i].value);
#ifdef RNG_TEST
*************** dump_seeds(int tbl)
*** 154,160 ****
#ifdef RNG_TEST
printf("%d(%ld):\t%ld\n", i, Seed[i].nCalls, Seed[i].value);
#else
! printf("%d:\t%lld\n", i, Seed[i].value);
#endif
return;
}
--- 153,159 ----
#ifdef RNG_TEST
printf("%d(%ld):\t%ld\n", i, Seed[i].nCalls, Seed[i].value);
#else
! printf("%d:\t%ld\n", i, Seed[i].value);
#endif
return;
}
*************** UnifInt(DSS_HUGE nLow, DSS_HUGE nHigh, l
*** 212,223 ****
if ((nHigh == MAX_LONG) && (nLow == 0))
{
! // Hyrise: was undefined behaviour, let's make it defined
! // dRange = DOUBLE_CAST (nHigh32 - nLow32 + 1);
! // nRange = nHigh32 - nLow32 + 1;
! dRange = DOUBLE_CAST(MIN_LONG);
! nRange = MIN_LONG;
!
}
else
{
--- 211,218 ----
if ((nHigh == MAX_LONG) && (nLow == 0))
{
! dRange = DOUBLE_CAST (nHigh32 - nLow32 + 1);
! nRange = nHigh32 - nLow32 + 1;
}
else
{
*************** UnifInt(DSS_HUGE nLow, DSS_HUGE nHigh, l
*** 233,289 ****
return (nLow + nTemp);
}
- seed_t Seed[MAX_STREAM + 1] =
- {
- {PART, 1, 0, 1}, /* P_MFG_SD 0 */
- {PART, 46831694, 0, 1}, /* P_BRND_SD 1 */
- {PART, 1841581359, 0, 1}, /* P_TYPE_SD 2 */
- {PART, 1193163244, 0, 1}, /* P_SIZE_SD 3 */
- {PART, 727633698, 0, 1}, /* P_CNTR_SD 4 */
- {NONE, 933588178, 0, 1}, /* text pregeneration 5 */
- {PART, 804159733, 0, 2}, /* P_CMNT_SD 6 */
- {PSUPP, 1671059989, 0, SUPP_PER_PART}, /* PS_QTY_SD 7 */
- {PSUPP, 1051288424, 0, SUPP_PER_PART}, /* PS_SCST_SD 8 */
- {PSUPP, 1961692154, 0, SUPP_PER_PART * 2}, /* PS_CMNT_SD 9 */
- {ORDER, 1227283347, 0, 1}, /* O_SUPP_SD 10 */
- {ORDER, 1171034773, 0, 1}, /* O_CLRK_SD 11 */
- {ORDER, 276090261, 0, 2}, /* O_CMNT_SD 12 */
- {ORDER, 1066728069, 0, 1}, /* O_ODATE_SD 13 */
- {LINE, 209208115, 0, O_LCNT_MAX}, /* L_QTY_SD 14 */
- {LINE, 554590007, 0, O_LCNT_MAX}, /* L_DCNT_SD 15 */
- {LINE, 721958466, 0, O_LCNT_MAX}, /* L_TAX_SD 16 */
- {LINE, 1371272478, 0, O_LCNT_MAX}, /* L_SHIP_SD 17 */
- {LINE, 675466456, 0, O_LCNT_MAX}, /* L_SMODE_SD 18 */
- {LINE, 1808217256, 0, O_LCNT_MAX}, /* L_PKEY_SD 19 */
- {LINE, 2095021727, 0, O_LCNT_MAX}, /* L_SKEY_SD 20 */
- {LINE, 1769349045, 0, O_LCNT_MAX}, /* L_SDTE_SD 21 */
- {LINE, 904914315, 0, O_LCNT_MAX}, /* L_CDTE_SD 22 */
- {LINE, 373135028, 0, O_LCNT_MAX}, /* L_RDTE_SD 23 */
- {LINE, 717419739, 0, O_LCNT_MAX}, /* L_RFLG_SD 24 */
- {LINE, 1095462486, 0, O_LCNT_MAX * 2}, /* L_CMNT_SD 25 */
- {CUST, 881155353, 0, 9}, /* C_ADDR_SD 26 */
- {CUST, 1489529863, 0, 1}, /* C_NTRG_SD 27 */
- {CUST, 1521138112, 0, 3}, /* C_PHNE_SD 28 */
- {CUST, 298370230, 0, 1}, /* C_ABAL_SD 29 */
- {CUST, 1140279430, 0, 1}, /* C_MSEG_SD 30 */
- {CUST, 1335826707, 0, 2}, /* C_CMNT_SD 31 */
- {SUPP, 706178559, 0, 9}, /* S_ADDR_SD 32 */
- {SUPP, 110356601, 0, 1}, /* S_NTRG_SD 33 */
- {SUPP, 884434366, 0, 3}, /* S_PHNE_SD 34 */
- {SUPP, 962338209, 0, 1}, /* S_ABAL_SD 35 */
- {SUPP, 1341315363, 0, 2}, /* S_CMNT_SD 36 */
- {PART, 709314158, 0, 92}, /* P_NAME_SD 37 */
- {ORDER, 591449447, 0, 1}, /* O_PRIO_SD 38 */
- {LINE, 431918286, 0, 1}, /* HVAR_SD 39 */
- {ORDER, 851767375, 0, 1}, /* O_CKEY_SD 40 */
- {NATION, 606179079, 0, 2}, /* N_CMNT_SD 41 */
- {REGION, 1500869201, 0, 2}, /* R_CMNT_SD 42 */
- {ORDER, 1434868289, 0, 1}, /* O_LCNT_SD 43 */
- {SUPP, 263032577, 0, 1}, /* BBB offset 44 */
- {SUPP, 753643799, 0, 1}, /* BBB type 45 */
- {SUPP, 202794285, 0, 1}, /* BBB comment 46 */
- {SUPP, 715851524, 0, 1} /* BBB junk 47 */
- };
- double dM = 2147483647.0;
\ No newline at end of file
--- 228,232 ----
diff -p third_party/tpch-dbgen/rnd.h ../tpch-dbgen/rnd.h
*** third_party/tpch-dbgen/rnd.h 2018-05-11 00:58:51.091846264 +0200
--- ../tpch-dbgen/rnd.h 2019-04-16 11:39:42.890690901 +0200
***************
*** 39,58 ****
DSS_HUGE NextRand PROTO((DSS_HUGE));
DSS_HUGE UnifInt PROTO((DSS_HUGE, DSS_HUGE, long));
! //static long nA = 16807; /* the multiplier */
! //static long nM = 2147483647;/* the modulus == 2^31 - 1 */
! //static long nQ = 127773; /* the quotient nM / nA */
! //static long nR = 2836; /* the remainder nM % nA */
! extern double dM;
/*
* macros to control RNG and assure reproducible multi-stream
* runs without the need for seed files. Keep track of invocations of RNG
* and always round-up to a known per-row boundary.
*/
! /*
* preferred solution, but not initializing correctly
*/
#define VSTR_MAX(len) (long)(len / 5 + (len % 5 == 0)?0:1 + 1)
! extern seed_t Seed[MAX_STREAM + 1];
\ No newline at end of file
--- 39,108 ----
DSS_HUGE NextRand PROTO((DSS_HUGE));
DSS_HUGE UnifInt PROTO((DSS_HUGE, DSS_HUGE, long));
! static long nA = 16807; /* the multiplier */
! static long nM = 2147483647;/* the modulus == 2^31 - 1 */
! static long nQ = 127773; /* the quotient nM / nA */
! static long nR = 2836; /* the remainder nM % nA */
! double dM = 2147483647.0;
/*
* macros to control RNG and assure reproducible multi-stream
* runs without the need for seed files. Keep track of invocations of RNG
* and always round-up to a known per-row boundary.
*/
! /*
* preferred solution, but not initializing correctly
*/
#define VSTR_MAX(len) (long)(len / 5 + (len % 5 == 0)?0:1 + 1)
! seed_t Seed[MAX_STREAM + 1] =
! {
! {PART, 1, 0, 1}, /* P_MFG_SD 0 */
! {PART, 46831694, 0, 1}, /* P_BRND_SD 1 */
! {PART, 1841581359, 0, 1}, /* P_TYPE_SD 2 */
! {PART, 1193163244, 0, 1}, /* P_SIZE_SD 3 */
! {PART, 727633698, 0, 1}, /* P_CNTR_SD 4 */
! {NONE, 933588178, 0, 1}, /* text pregeneration 5 */
! {PART, 804159733, 0, 2}, /* P_CMNT_SD 6 */
! {PSUPP, 1671059989, 0, SUPP_PER_PART}, /* PS_QTY_SD 7 */
! {PSUPP, 1051288424, 0, SUPP_PER_PART}, /* PS_SCST_SD 8 */
! {PSUPP, 1961692154, 0, SUPP_PER_PART * 2}, /* PS_CMNT_SD 9 */
! {ORDER, 1227283347, 0, 1}, /* O_SUPP_SD 10 */
! {ORDER, 1171034773, 0, 1}, /* O_CLRK_SD 11 */
! {ORDER, 276090261, 0, 2}, /* O_CMNT_SD 12 */
! {ORDER, 1066728069, 0, 1}, /* O_ODATE_SD 13 */
! {LINE, 209208115, 0, O_LCNT_MAX}, /* L_QTY_SD 14 */
! {LINE, 554590007, 0, O_LCNT_MAX}, /* L_DCNT_SD 15 */
! {LINE, 721958466, 0, O_LCNT_MAX}, /* L_TAX_SD 16 */
! {LINE, 1371272478, 0, O_LCNT_MAX}, /* L_SHIP_SD 17 */
! {LINE, 675466456, 0, O_LCNT_MAX}, /* L_SMODE_SD 18 */
! {LINE, 1808217256, 0, O_LCNT_MAX}, /* L_PKEY_SD 19 */
! {LINE, 2095021727, 0, O_LCNT_MAX}, /* L_SKEY_SD 20 */
! {LINE, 1769349045, 0, O_LCNT_MAX}, /* L_SDTE_SD 21 */
! {LINE, 904914315, 0, O_LCNT_MAX}, /* L_CDTE_SD 22 */
! {LINE, 373135028, 0, O_LCNT_MAX}, /* L_RDTE_SD 23 */
! {LINE, 717419739, 0, O_LCNT_MAX}, /* L_RFLG_SD 24 */
! {LINE, 1095462486, 0, O_LCNT_MAX * 2}, /* L_CMNT_SD 25 */
! {CUST, 881155353, 0, 9}, /* C_ADDR_SD 26 */
! {CUST, 1489529863, 0, 1}, /* C_NTRG_SD 27 */
! {CUST, 1521138112, 0, 3}, /* C_PHNE_SD 28 */
! {CUST, 298370230, 0, 1}, /* C_ABAL_SD 29 */
! {CUST, 1140279430, 0, 1}, /* C_MSEG_SD 30 */
! {CUST, 1335826707, 0, 2}, /* C_CMNT_SD 31 */
! {SUPP, 706178559, 0, 9}, /* S_ADDR_SD 32 */
! {SUPP, 110356601, 0, 1}, /* S_NTRG_SD 33 */
! {SUPP, 884434366, 0, 3}, /* S_PHNE_SD 34 */
! {SUPP, 962338209, 0, 1}, /* S_ABAL_SD 35 */
! {SUPP, 1341315363, 0, 2}, /* S_CMNT_SD 36 */
! {PART, 709314158, 0, 92}, /* P_NAME_SD 37 */
! {ORDER, 591449447, 0, 1}, /* O_PRIO_SD 38 */
! {LINE, 431918286, 0, 1}, /* HVAR_SD 39 */
! {ORDER, 851767375, 0, 1}, /* O_CKEY_SD 40 */
! {NATION, 606179079, 0, 2}, /* N_CMNT_SD 41 */
! {REGION, 1500869201, 0, 2}, /* R_CMNT_SD 42 */
! {ORDER, 1434868289, 0, 1}, /* O_LCNT_SD 43 */
! {SUPP, 263032577, 0, 1}, /* BBB offset 44 */
! {SUPP, 753643799, 0, 1}, /* BBB type 45 */
! {SUPP, 202794285, 0, 1}, /* BBB comment 46 */
! {SUPP, 715851524, 0, 1} /* BBB junk 47 */
! };
Only in ../tpch-dbgen/: rnd.o
Only in ../tpch-dbgen/: rng64.o
diff -p third_party/tpch-dbgen/speed_seed.c ../tpch-dbgen/speed_seed.c
*** third_party/tpch-dbgen/speed_seed.c 2018-05-11 00:58:51.091846264 +0200
--- ../tpch-dbgen/speed_seed.c 2019-04-16 11:39:42.890690901 +0200
*************** extern seed_t Seed[];
*** 66,72 ****
void fakeVStr(int nAvg, long nSeed, DSS_HUGE nCount);
void NthElement (DSS_HUGE N, DSS_HUGE *StartSeed);
- static char lnoise[4] = {'|', '/', '-', '\\' };
void
advanceStream(int nStream, DSS_HUGE nCalls, int bUse64Bit)
--- 66,71 ----
Only in ../tpch-dbgen/: speed_seed.o
Only in ../tpch-dbgen/: supplier.tbl
Only in ../tpch-dbgen/: tests
diff -p third_party/tpch-dbgen/text.c ../tpch-dbgen/text.c
*** third_party/tpch-dbgen/text.c 2019-04-16 17:17:08.599503486 +0200
--- ../tpch-dbgen/text.c 2019-04-16 11:39:42.890690901 +0200
*************** next_token: /* I hate goto's, but can't
*** 253,260 ****
case 'P':
i = pick_str(&prepositions, sd, dest);
len = (int)strlen(DIST_MEMBER(&prepositions, i));
! // HYRISE: Avoid redundantly figuring out the string length - use memcpy over strcpy
! memcpy((dest + len), " the ", 5);
len += 5;
len += txt_np(dest + len, sd);
break;
--- 253,259 ----
case 'P':
i = pick_str(&prepositions, sd, dest);
len = (int)strlen(DIST_MEMBER(&prepositions, i));
! strcpy((dest + len), " the ");
len += 5;
len += txt_np(dest + len, sd);
break;
*************** dbg_text(char *tgt, int min, int max, in
*** 317,324 ****
needed = TEXT_POOL_SIZE - wordlen;
if (needed >= (s_len + 1)) /* need the entire sentence */
{
! // HYRISE: Avoid redundantly figuring out the string length - use memcpy over strcpy
! memcpy(cp, sentence, s_len);
cp += s_len;
wordlen += s_len + 1;
*(cp++) = ' ';
--- 316,322 ----
needed = TEXT_POOL_SIZE - wordlen;
if (needed >= (s_len + 1)) /* need the entire sentence */
{
! strcpy(cp, sentence);
cp += s_len;
wordlen += s_len + 1;
*(cp++) = ' ';
*************** dbg_text(char *tgt, int min, int max, in
*** 326,333 ****
else /* chop the new sentence off to match the length target */
{
sentence[needed] = '\0';
! // HYRISE: Avoid redundantly figuring out the string length - use memcpy over strcpy
! memcpy(cp, sentence, needed);
wordlen += needed;
cp += needed;
}
--- 324,330 ----
else /* chop the new sentence off to match the length target */
{
sentence[needed] = '\0';
! strcpy(cp, sentence);
wordlen += needed;
cp += needed;
}
*************** dbg_text(char *tgt, int min, int max, in
*** 340,347 ****
RANDOM(hgOffset, 0, TEXT_POOL_SIZE - max, sd);
RANDOM(hgLength, min, max, sd);
! // HYRISE: Avoid redundantly figuring out the string length - use memcpy over strcpy
! memcpy(&tgt[0], &szTextPool[hgOffset], (int)hgLength);
tgt[hgLength] = '\0';
return;
--- 337,343 ----
RANDOM(hgOffset, 0, TEXT_POOL_SIZE - max, sd);
RANDOM(hgLength, min, max, sd);
! strncpy(&tgt[0], &szTextPool[hgOffset], (int)hgLength);
tgt[hgLength] = '\0';
return;
Only in ../tpch-dbgen/: text.o
Only in ../tpch-dbgen/: tpcd.h
Only in third_party/tpch-dbgen: tpch_dbgen.c
Only in third_party/tpch-dbgen: tpch_dbgen.h
Only in ../tpch-dbgen/: tpch.dsw
Only in ../tpch-dbgen/: tpch.sln
Only in ../tpch-dbgen/: tpch.vcproj
Only in ../tpch-dbgen/: variants
Only in ../tpch-dbgen/: varsub.c
Only in ../tpch-dbgen/: varsub.o
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment