along/secondo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
dfd1e745480fabd88139d2804acb90d5b6dc055e
secondo/Tools/Generators/TPC-D/Bugs

550 lines
17 KiB
Plaintext
Raw Normal View History

firs commit
2026-01-23 17:03:45 +08:00
{ SCCS: @(#)BUGS 9.1.1.9 5/1/96 11:55:26 }
The following is a list of the various DBGEN/QGEN bugs that have been
and are being fixed. Each entry is of the form:
Problem #xx: -- a brief description of the issue or problem,
followed by a more detailed explanation, if needed
SEVERITY: -- classification of the bug or issue
SPEC FIX: -- details of any change to the spec
DBGEN FIX: -- details of any change needed to QGEN/DBGEN
ANSWER SETS: -- any effect on answer sets
STATUS: -- status of resolution
WORKAROUND: -- temporary fix, if available
HELP NEEDED: -- any work/assistance required
AUDITORS NOTIFIED: -- date auditors were notified, if appropriate
OPENED AGAINST: -- date and effected versions
CLOSED IN: -- date and fixed version
Open BUGS
==========
Problem #33: Parallel load doesn't work under NT
Feature Requests
=================
Problem #9: would like to include answer set formatting in query templates
Problem #37: need way to validate DBGEN without large storage requriement
Complete Bug List
==================
Problem #1: Q10 returns no rows
Since orders can only be returned (l_returnflag = 'R') after they
have been received, and can't be received in the future, the
number of permissible orders for query 10 tails off early in
1995. If you are lucky enough to get a parameter substitution
after February '95 (allowed in 2.12.3), things can go "quickly".
SEVERITY:
SPEC FIX: replace 2.12.3 (1) with "DATE is the first day in a
rundomly selected month between the first month of 1993 and the
last month of 1994"
DBGEN FIX: change permisible substitution range for query 10,
parameter 1
ANSWER SETS: not effected.
STATUS: closed
WORKAROUND: use a different seed for qgen parameter substitution
HELP NEEDED:
AUDITORS NOTIFIED:
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1 (dbgen and qgen)
Problem #2: parallelism in load to gen differing data sets
the parallel load code was based on extensible data sets; since
each "extension" made an assumption of scale factor, the data
could end up clustered. Further, since the RNG is
self-modifying, different numbers of extension led to different
final data sets.
SEVERITY:
SPEC FIX: none.
DBGEN FIX: remove -E(xtensible) option and implement pure parallel
load with a known scale factor; rebuild seed files
ANSWER SETS: not effected. (parallelism not implemented for SF <= 1)
STATUS: closed
WORKAROUND: don't use the parallel load (-C) option to DBGEN
HELP NEEDED: testers needed.
AUDITORS NOTIFIED: yes.
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #3: some arithmetic tends to overflow at large SF
retailprice tends to SF/10 as SF increases. this can lead to
data corruption in extendedprice and aggregate calculations
SEVERITY:
SPEC FIX: will need rework of 1.3 wrt retailprice calculation
DBGEN FIX: modification to second term of rpb_routine() calcuation
to limit contibution of second term to the maximum seen at
SF=.1
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: code retail/extended price calculations as long long;
build smaller data sets
HELP NEEDED:
AUDITORS NOTIFIED:
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #4: dbgen not ported to NT
SEVERITY:
SPEC FIX: none
DBGEN FIX: need to roll in changes supplied by IBM
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: N/A
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.1.0
Problem #5: QGEN seed init inconsistent
A prior fix assured that parameter values were query order
independent when a seed was provided on the command line. need
to make this true when no seed is provided
SEVERITY:
SPEC FIX: none
DBGEN FIX: rework seed init loop in qgen.c
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: supply seeds on command line
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #6: command line options with abutting arguments mishandled
SEVERITY:
SPEC FIX: none
DBGEN FIX: minor fix to getopt routine in bm_utils.c
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: separate options and arguments with a space
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #7: '-O f' asking for new file names twice
SEVERITY:
SPEC FIX: none
DBGEN FIX: rework of set_files() in driver.c
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: none
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #8: Seed generation taking too long
SEVERITY:
SPEC FIX: N/A
DBGEN FIX: implement "skip and trudge" as discussed
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: none
HELP NEEDED:
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #9: would like to include answer set formatting in query templates
SEVERITY: feature request
SPEC FIX: none
DBGEN FIX: additional flag in qgen()
ANSWER SETS: not effected
STATUS:
WORKAROUND: N/A
HELP NEEDED: asked for reproduction info 25 Oct 95
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN:
Problem #10: need to re-introduce ability to do incremental, flat file builds
SEVERITY: feature request
SPEC FIX: none
DBGEN FIX: add -S(tep) option to build one of many partial data sets
ANSWER SETS: not effected
STATUS: closed
WORKAROUND: N/A
HELP NEEDED:
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0
CLOSED IN: 1.0.1
Problem #11: Row count for first delete at 10/100 is incorrect
SEVERITY: Error
SPEC FIX: None
DBGEN FIX:
ANSWER SETS: No Effect
STATUS: OPEN
WORKAROUND: hand edit of first delete file
HELP NEEDED:
AUDITORS NOTIFIED: No
OPENED AGAINST: 1.0.1
CLOSED IN:
Problem #12: Bad default rowcount generated for query 17
SEVERITY: Error
SPEC FIX: None
DBGEN FIX: corrected rowcnt[] entries to be 1-based
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: hand edit query or add explicit row count to template
HELP NEEDED:
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.0
CLOSED IN: 1.1.0
Problem #13: Bad expansion of SET_OUTPUT for Teradata
SEVERITY: Error
SPEC FIX: N/A
DBGEN FIX: new macro in tpcd.h
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: Hand edit query or hardcode output directive in templates
HELP NEEDED:
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.1
CLOSED IN: 1.1.0
Problem #14: Badly formed range deletes
SEVERITY: Error
SPEC FIX: N/A
DBGEN FIX: TBD
ANSWER SETS: N/A
STATUS: OPEN
WORKAROUND: hand edit delete files
HELP NEEDED: asked for reproduction info 25 Oct 95
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.1
CLOSED IN:
Problem #15: in a multi-stage load, parent tables are not properly named
when parent and child are build simultaneously
SEVERITY: Error
SPEC FIX: N/A
DBGEN FIX: reworked tdef[].name in pr_X_Y routines for master/detail
tables
ANSWER SETS: N/A
STATUS: open
WORKAROUND: Build master/detail tables separately
HELP NEEDED:
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.1
CLOSED IN: 1.1.0
Problem #16: update generation at large scale factors produced the wrong number
of rows due to overflow of 32-bit integer
SEVERITY: BUG
SPEC FIX: N/A
DBGEN FIX: corrected order of operations in row count calcuation in
driver.c
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: use 64 bit integers
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.1
CLOSED IN: 1.1.0
Problem #17: comment fields may be truncated when using columnar output, due to
rounding/truncation in the length calculation
SEVERITY: BUG
SPEC FIX: N/A
DBGEN FIX: add ceil() calls around all PR_VSTR() calls in print.c
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.0.1
CLOSED IN: 1.1.0
Problem #18: the output format for identifier fields in columnar output is
unneccessarily large, and is inconsistant
SEVERITY: minor
SPEC FIX: N/A
DBGEN FIX: revised PR_BCD2 macro
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: avoid columnar output, or rework macro
HELP NEEDED: none
AUDITORS NOTIFIED: no
OPENED AGAINST: 1.1.0
OPENED BY: jenn@torolab2.vnet.ibm.com
CLOSED IN: 1.1.0A
CLOSED BY: jms@informix.com
Problem #19: the case statement used to decipher substitution points in the
query template allowed extraneous :'s to re-initialize the
parameter substitution
SEVERITY: bug
SPEC FIX: N/A
DBGEN FIX: rework flag switch in qgen.c to explicitly call out numerics
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: be sure that there are no "unknown" flags in the template
HELP NEEDED: none
AUDITORS NOTIFIED: yes
OPENED AGAINST: 1.0.1
OPENED BY: jenn@torolab2.vnet.ibm.com
CLOSED IN: 1.1.0A
CLOSED BY: jms@informix.com
Problem #20: parameter substitution values were not effected by small changes
in seed values
SEVERITY: bug
SPEC FIX: N/A
DBGEN FIX: add UnifInt() calls to RNG init in qgen.c
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: be sure seed values provide sufficient randomness in EQT
HELP NEEDED: none
AUDITORS NOTIFIED: yes
OPENED AGAINST: 1.1.0
OPENED BY: alain_crolotte@elsegundoca.attgis.com
CLOSED IN: 1.1.0B
CLOSED BY: jms@informix.com
Problem #21: parameter logging doesn't properly handle the variable length of
the substitution list
SEVERITY: bug
SPEC FIX: N/A
DBGEN FIX: assure null termination of param list and bound the output
loop that logs parameter usage
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: none
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0B
OPENED BY:
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #22: parameter output for Q11 can overflow default formatting at very
large volumes
SEVERITY: bug
SPEC FIX: N/A
DBGEN FIX: expand format string to %11.10f
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: hand code queries for large volumes
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0B
OPENED BY: francois@ip.com
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #23: typos in variant 14c
SEVERITY:
SPEC FIX: N/A
DBGEN FIX: corrected query template
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: none
HELP NEEDED: none
AUDITORS NOTIFIED: no
OPENED AGAINST: 1.1.0B
OPENED BY: francois@ip.com
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #24: macro PR_DATE was hard-coded to print t->alpha even though a
target was passed in as a parameter
SEVERITY: minor
SPEC FIX: N/A
SOURCE FIX: re-worked macro to properly use its arguments
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: none
HELP NEEDED: none
AUDITORS NOTIFIED: no
OPENED AGAINST: 1.1.0A
OPENED BY: Robert.Lane@eng.sun.com
CLOSED IN: dbgen 1.1.0B
CLOSED BY: jms@informix.com
Problem #25: typos in variant 10a
SEVERITY:
SPEC FIX: N/A
DBGEN FIX: corrected query template
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: none
HELP NEEDED: none
AUDITORS NOTIFIED: no
OPENED AGAINST: 1.1.0B
OPENED BY: francois@ip.com
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #26: the version numbers for QGEN and DBGEN do not match
SEVERITY: minor
SPEC FIX: N/A
SOURCE FIX: unified version numbers starting with 1.1.0C
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: none
AUDITORS NOTIFIED: no
OPENED AGAINST: 1.1.0B (or 1.1.0C, depending)
OPENED BY: Robert.Lane@eng.sun.com
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #27: correcting typos in 7, 9, 13
SEVERITY: minor
SPEC FIX: N/A
SOURCE FIX: fixed them
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0C (pre-release)
OPENED BY: tblank@vnet.ibm.com
CLOSED IN: 1.1.0C
CLOSED BY: jms@informix.com
Problem #28: Seed generation fails with SF > 1000 due to 32 bit integer
arithmetic used to verify "divisible-ness" of data set
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: TBD
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0C
OPENED BY: alain_colotte@elsegundoca.ncr.com
CLOSED IN: 1.3.0
CLOSED BY: jms@gradientsystems.com
Problem #29: Compile time errors on Solaris 2.5.1 and SunOS
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: Solaris fixed by renaming lineitem field from extended to
eprice; SunOS problem documented in Porting.Notes
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: N/A
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0D
OPENED BY: jms@informix.com
CLOSED IN: 1.2.0
CLOSED BY: jms@informix.com
Problem #30: Cryptic comments in dists.dss
SEVERITY: flaw
SPEC FIX: N/A
SOURCE FIX: Cleaned up the comments in the file
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: N/A
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.0
OPENED BY: francois@sizing.com
CLOSED IN: 1.2.3 ALPHA 1
CLOSED BY: jms@informix.com
Problem #31: Inconsistant handling of fopen() failures
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: introduced OPEN_CHECK macro (defined in dss.h)
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.0
OPENED BY: schiefer@ca.ibm.com
CLOSED IN: 1.3.0
CLOSED BY: jms@gradientsystems.com
Problem #32: Path separators were hard-coded
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: introduced PATH_SEP in config.h
ANSWER SETS: N/A
STATUS: closed
WORKAROUND: N/A
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.0
OPENED BY:
CLOSED IN: 1.3.0
CLOSED BY: jms@gradientsystems.com
Problem #33: Parallel load doesn't work under NT
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX:
ANSWER SETS: N/A
STATUS: OPEN
WORKAROUND: use -S option to build each step independently
HELP NEEDED: none
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.1.0
OPENED BY:
CLOSED IN:
CLOSED BY:
Problem #34: P_NAME not properly populated
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: Corrected color selection logic in agg_str()
ANSWER SETS: NFI for 1.x since it effect answer sets
STATUS: Closed
WORKAROUND: N/A
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.3
OPENED BY: schiefer@ca.ibm.com
CLOSED IN: 2.0.0
CLOSED BY: jms@gradientsystems.com
Problem #35: mk_sparse() returning bad orderkeys
SEVERITY: bug
SPEC FIX: N/A
SOURCE FIX: corrected logic in mk_sparse() and bcd2_bin()
ANSWER SETS: N/A
STATUS: Closed
WORKAROUND: N/A
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.3.0
OPENED BY: jennc@ca.ibm.com
CLOSED IN: 1.3.1
CLOSED BY: jms@gradientsystems.com
Problem #36: a_rnd() doesn't mask properly, uses small 'alphabet'
SEVERITY: bug
SPEC FIX: Corrected 4.2.2.6 to reflect 64 character set
SOURCE FIX: changed mask in a_rnd() from 067 to 077
ANSWER SETS: NFI for 1.x since answers would be effected
STATUS: Closed
WORKAROUND: N/A
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.3
OPENED BY: pek@elsegundoca.ncr.com
CLOSED IN: 2.0.0
CLOSED BY: jms@gradientsystems.com
Problem #37: need way to validate DBGEN without large storage requriement
SEVERITY: Feature Request
SPEC FIX: N/A
SOURCE FIX: Provide vrf_xxx routine to generate checksums
ANSWER SETS: N/A
STATUS: Open
WORKAROUND: N/A
HELP NEEDED: None
AUDITORS NOTIFIED: N/A
OPENED AGAINST: 1.2.3
OPENED BY: jms@gradientsystems.com
CLOSED IN:
CLOSED BY:
Reference in New Issue Copy Permalink