Tom Lane [Mon, 23 Aug 1999 23:48:39 +0000 (23:48 +0000)]
Remove bogus code in oper_exact --- if it didn't find an exact
match then it tried for a self-commutative operator with the reversed input
data types. This is pretty silly; there could never be such an operator,
except maybe in binary-compatible-type scenarios, and we have oper_inexact
for that. Besides which, the oprsanity regress test would complain about
such an operator. Remove nonfunctional code and simplify routine calling
convention accordingly.
Tom Lane [Sun, 22 Aug 1999 23:56:45 +0000 (23:56 +0000)]
Further sort-order twiddling in optimizer: be smart about
case where ORDER BY and GROUP BY request the same sort order.
Tom Lane [Sun, 22 Aug 1999 20:37:24 +0000 (20:37 +0000)]
Un-break code generated by ECPGRelease production.
Tom Lane [Sun, 22 Aug 1999 20:15:04 +0000 (20:15 +0000)]
Further planner/optimizer cleanups. Move all set_tlist_references
and fix_opids processing to a single recursive pass over the plan tree
executed at the very tail end of planning, rather than haphazardly here
and there at different places. Now that tlist Vars do not get modified
until the very end, it's possible to get rid of the klugy var_equal and
match_varid partial-matching routines, and just use plain equal()
throughout the optimizer. This is a step towards allowing merge and
hash joins to be done on expressions instead of only Vars ...
Tom Lane [Sat, 21 Aug 1999 03:49:17 +0000 (03:49 +0000)]
Major revision of sort-node handling: push knowledge of query
sort order down into planner, instead of handling it only at the very top
level of the planner. This fixes many things. An explicit sort is now
avoided if there is a cheaper alternative (typically an indexscan) not
only for ORDER BY, but also for the internal sort of GROUP BY. It works
even when there is no other reason (such as a WHERE condition) to consider
the indexscan. It works for indexes on functions. It works for indexes
on functions, backwards. It's just so cool...
CAUTION: I have changed the representation of SortClause nodes, therefore
THIS UPDATE BREAKS STORED RULES. You will need to initdb.
Tom Lane [Sat, 21 Aug 1999 03:06:58 +0000 (03:06 +0000)]
Cleanups for int8: guard against null inputs in comparison
operators (and some other places), fix rangechecks in int8 to int4
conversion (same problem we recently figured out in pg_atoi).
Tom Lane [Sat, 21 Aug 1999 00:56:18 +0000 (00:56 +0000)]
Ooops ... I had left some test coding in selfuncs.c that
failed on 'field < textconstant' ...
Bruce Momjian [Wed, 18 Aug 1999 15:11:52 +0000 (15:11 +0000)]
Update TODO list.
Tatsuo Ishii [Wed, 18 Aug 1999 13:04:45 +0000 (13:04 +0000)]
Old multi-byte bug. Forgot to rename #ifdef MB to #ifdef MULTIBYTE
Now SET NAMES working again...
Tom Lane [Wed, 18 Aug 1999 04:15:16 +0000 (04:15 +0000)]
Remove extraneous SeqScan node that make_noname was inserting
above a Sort or Materialize node. As far as I can tell, the only place
that actually needed that was set_tlist_references, which was being lazy
about checking to see if it had a noname node to fix or not...
Tom Lane [Tue, 17 Aug 1999 21:21:22 +0000 (21:21 +0000)]
Add script that runs the regression tests with all valid
combinations of query-plan-type backend options. Good for testing
planner/optimizer. Tedious, though.
Bruce Momjian [Tue, 17 Aug 1999 17:31:11 +0000 (17:31 +0000)]
Add missing docs semicolon.
Tom Lane [Mon, 16 Aug 1999 23:47:23 +0000 (23:47 +0000)]
EXPLAIN didn't know about 'Materialize' plan nodes.
Tom Lane [Mon, 16 Aug 1999 23:07:20 +0000 (23:07 +0000)]
Assign sort keys properly when there are duplicate entries in
pathkey list --- corrects misbehavior seen with multiple mergejoin clauses
mentioning same variable.
Bruce Momjian [Mon, 16 Aug 1999 20:32:34 +0000 (20:32 +0000)]
Typo fix
Bruce Momjian [Mon, 16 Aug 1999 20:27:19 +0000 (20:27 +0000)]
I've sent 3 mails to pgsql-patches. There are two files, one for doc
and
for src/data directories, and one minor patch for doc/README.locale.
Please apply.
Oleg.
Bruce Momjian [Mon, 16 Aug 1999 20:12:12 +0000 (20:12 +0000)]
Fix for perl5 on BSD/OS.
Bruce Momjian [Mon, 16 Aug 1999 19:59:25 +0000 (19:59 +0000)]
Change >& to > and 2>&1.
Bruce Momjian [Mon, 16 Aug 1999 19:57:21 +0000 (19:57 +0000)]
Allow BSD yacc and bison to compile pl code.
Tom Lane [Mon, 16 Aug 1999 02:17:58 +0000 (02:17 +0000)]
Major planner/optimizer revision: get rid of PathOrder node type,
store all ordering information in pathkeys lists (which are now lists of
lists of PathKeyItem nodes, not just lists of lists of vars). This was
a big win --- the code is smaller and IMHO more understandable than it
was, even though it handles more cases. I believe the node changes will
not force an initdb for anyone; planner nodes don't show up in stored
rules.
Tom Lane [Mon, 16 Aug 1999 02:10:13 +0000 (02:10 +0000)]
Small updates to #include lists for pending optimizer checkin.
Tom Lane [Mon, 16 Aug 1999 02:08:59 +0000 (02:08 +0000)]
Move funcid_get_rettype() to lsyscache.
Tom Lane [Mon, 16 Aug 1999 02:06:25 +0000 (02:06 +0000)]
Add get_func_rettype() to general-use lsyscache routines,
since it's now needed in both optimizer and parser.
Tatsuo Ishii [Mon, 16 Aug 1999 02:04:05 +0000 (02:04 +0000)]
Fix for Win32 making problem with MB enabled.
Patches created by Hiroki Kataoka.
Thomas G. Lockhart [Sun, 15 Aug 1999 06:46:49 +0000 (06:46 +0000)]
Repair the check for redundant UNIQUE and PRIMARY KEY indices.
Also, improve it so that it checks for multi-column constraints.
Thanks to Mark Dalphin for reporting the problem.
Tom Lane [Sat, 14 Aug 1999 19:30:21 +0000 (19:30 +0000)]
LispUnion fix changes output ordering for inheritance tests.
Tom Lane [Sat, 14 Aug 1999 19:29:35 +0000 (19:29 +0000)]
LispUnion routine didn't generate a proper union: anytime
l2 contained more than one entry, there would be duplicates in the output
list. Miscellaneous code beautification in other routines, too.
Bruce Momjian [Fri, 13 Aug 1999 16:56:04 +0000 (16:56 +0000)]
Update TODO list.
Tom Lane [Fri, 13 Aug 1999 01:17:16 +0000 (01:17 +0000)]
Update comments about pathkeys.
Tom Lane [Thu, 12 Aug 1999 04:32:54 +0000 (04:32 +0000)]
Clean up optimizer's handling of indexscan quals that need to be
commuted (ie, the index var appears on the right). These are now handled
the same way as merge and hash join quals that need to be commuted: the
actual reversing of the clause only happens if we actually choose the path
and generate a plan from it. Furthermore, the clause is only reversed in
the 'indexqual' field of the plan, not in the 'indxqualorig' field. This
allows the clause to still be recognized and removed from qpquals of upper
level join plans. Also, simplify and generalize match_clause_to_indexkey;
now it recognizes binary-compatible indexes for join as well as restriction
clauses.
Tom Lane [Thu, 12 Aug 1999 00:42:43 +0000 (00:42 +0000)]
Add commentary to show that even though ExecInitIndexScan()
contains much code that looks like it will handle indexquals with the index
key on either side of the operator, in fact indexquals must have the index
key on the left because of limitations of the ScanKey machinery. Perhaps
someone will be motivated to fix that someday...
Bruce Momjian [Tue, 10 Aug 1999 16:56:54 +0000 (16:56 +0000)]
Update TODO list.
Tom Lane [Tue, 10 Aug 1999 03:00:15 +0000 (03:00 +0000)]
Minor cleanups and code beautification; eliminate some
routines that are now dead code.
Tom Lane [Tue, 10 Aug 1999 02:58:56 +0000 (02:58 +0000)]
Revise create_nestloop_node's handling of inner indexscan to
work under a wider range of scenarios than it did --- it formerly did not
handle a multi-pass inner scan, nor cases in which the inner scan's
indxqualorig or non-index qual contained outer var references. I am not
sure that these limitations could be hit in the existing optimizer, but
they need to be fixed for future expansion.
Bruce Momjian [Mon, 9 Aug 1999 06:20:27 +0000 (06:20 +0000)]
> > Prevent sorting if result is already sorted
> >
> > was implemented by Jan Wieck.
> > His work is for ascending order cases.
> >
> > Here is a patch to prevent sorting also in descending
> > order cases.
> > Because I had already changed _bt_first() to position
> > backward correctly before v6.5,this patch would work.
> >
Hiroshi Inoue
[email protected]
Tom Lane [Mon, 9 Aug 1999 05:34:13 +0000 (05:34 +0000)]
Clean up tlist.c tree-walking routines with
expression_tree_mutator.
Tom Lane [Mon, 9 Aug 1999 03:16:47 +0000 (03:16 +0000)]
Store -1 in attdisbursion to signal 'no duplicates in column'.
Centralize att_disbursion readout logic.
Tom Lane [Mon, 9 Aug 1999 03:13:31 +0000 (03:13 +0000)]
Move get_attdisbursion to lsyscache. Clean up get_typdefault.
Tom Lane [Mon, 9 Aug 1999 02:45:56 +0000 (02:45 +0000)]
Update comments about attdisbursion. NO code change.
Vadim B. Mikheev [Mon, 9 Aug 1999 01:39:19 +0000 (01:39 +0000)]
Re-use free space on index pages with duplicates.
Tom Lane [Mon, 9 Aug 1999 01:01:42 +0000 (01:01 +0000)]
Rewrite fix_indxqual_references, which was entirely bogus for
multi-scan indexscan plans; it tried to use the same table-to-index
attribute mapping for all the scans, even if they used different indexes.
It would klugily work as long as OR indexquals never used multikey indexes,
but that's not likely to hold up much longer...
Tom Lane [Mon, 9 Aug 1999 00:56:05 +0000 (00:56 +0000)]
Clean up routines in setrefs.c by replacing individual tree
walking logic with expression_tree_walker/mutator calls.
Tom Lane [Mon, 9 Aug 1999 00:51:26 +0000 (00:51 +0000)]
Create a standardized expression_tree_mutator support routine
to go along with expression_tree_walker. (_walker is not suitable for
routines that need to alter the tree structure significantly.) Other minor
cleanups in clauses.c.
Bruce Momjian [Mon, 9 Aug 1999 00:08:52 +0000 (00:08 +0000)]
replace #Include .
Tom Lane [Sun, 8 Aug 1999 20:12:52 +0000 (20:12 +0000)]
Fix nbtree's failure to clear BTScans list during xact abort.
Also, move responsibility for calling vc_abort into main xact.c list of
things-to-call-at-abort. What in the world was it doing down inside of
TransactionIdAbort()?
Tom Lane [Sun, 8 Aug 1999 17:13:10 +0000 (17:13 +0000)]
For a unique-key attribute (no duplicate values), vacuum analyze
was recording a disbursion of 0, not the correct value 1/numberOfRows.
Thomas G. Lockhart [Sun, 8 Aug 1999 15:21:50 +0000 (15:21 +0000)]
Remove old man pages. All info is in new man pages or other docs.
Thomas G. Lockhart [Sun, 8 Aug 1999 15:10:21 +0000 (15:10 +0000)]
Use sgml/Makefile to generate man sections, so just "make man" there.
Thomas G. Lockhart [Sun, 8 Aug 1999 15:09:04 +0000 (15:09 +0000)]
Switch from old man pages to new ones generated from sgml.
Thomas G. Lockhart [Sun, 8 Aug 1999 15:05:22 +0000 (15:05 +0000)]
Remove emacs properties block from end of file.
Thomas G. Lockhart [Sun, 8 Aug 1999 04:21:33 +0000 (04:21 +0000)]
Update references to char2 type by using char(2).
Thanks to Garr Updegraff for the tip.
Thomas G. Lockhart [Sun, 8 Aug 1999 04:18:29 +0000 (04:18 +0000)]
Allow nsgmls to validate input by defining the document type.
Use refentry.sgml as a generic header for man pages through nsgmls.
Bruce Momjian [Sat, 7 Aug 1999 04:25:02 +0000 (04:25 +0000)]
Fix so BSD yacc works on file.
Thomas G. Lockhart [Fri, 6 Aug 1999 13:50:31 +0000 (13:50 +0000)]
Fix markup to allow clean man pages.
Add new pages for programming languages and tcl shells.
Thomas G. Lockhart [Fri, 6 Aug 1999 13:48:38 +0000 (13:48 +0000)]
Fix cross-reference markup so that only the *title* of the Operators
chapter is included, not the chapter itself.
Thanks to Evelyn Mitchell for pointing it out.
Thomas G. Lockhart [Fri, 6 Aug 1999 13:46:17 +0000 (13:46 +0000)]
Remove explicit references to ref/ path in file names; use vpath instead.
Fix rules for man pages to ensure double-pass to get cross references.
Add a few new man pages.
Thomas G. Lockhart [Fri, 6 Aug 1999 13:43:42 +0000 (13:43 +0000)]
Try to clarify characteristics of the SERIAL type.
Fix source indenting, which does not affect output.
Note: still need docs on NUMERIC and DECIMAL
(and let's not talk about regression tests :()
Tom Lane [Fri, 6 Aug 1999 04:00:17 +0000 (04:00 +0000)]
Revise generation of hashjoin paths: generate one path per
hashjoinable clause, not one path for a randomly-chosen element of each
set of clauses with the same join operator. That is, if you wrote
SELECT ... WHERE t1.f1 = t2.f2 and t1.f3 = t2.f4,
and both '=' ops were the same opcode (say, all four fields are int4),
then the system would either consider hashing on f1=f2 or on f3=f4,
but it would *not* consider both possibilities. Boo hiss.
Also, revise estimation of hashjoin costs to include a penalty when the
inner join var has a high disbursion --- ie, the most common value is
pretty common. This tends to lead to badly skewed hash bucket occupancy
and way more comparisons than you'd expect on average.
I imagine that the cost calculation still needs tweaking, but at least
it generates a more reasonable plan than before on George Young's example.
Tom Lane [Thu, 5 Aug 1999 02:35:29 +0000 (02:35 +0000)]
parse_coerce fix allows early simplification of one
constant-coercion expression in the rules test.
Tom Lane [Thu, 5 Aug 1999 02:33:54 +0000 (02:33 +0000)]
Revise parse_coerce() to handle coercion of int and float
constants, not only string constants, at parse time. Get rid of
parser_typecast2(), which is bogus and redundant...
Bruce Momjian [Thu, 5 Aug 1999 01:53:35 +0000 (01:53 +0000)]
Update TODO list.
Tom Lane [Tue, 3 Aug 1999 00:09:32 +0000 (00:09 +0000)]
Fix ELF test so it doesn't spit up on all non-ELF systems...
use Autoconf-approved method of testing for predefined symbols, and move
it down to where we know what compiler to run and how to run it.
Tom Lane [Mon, 2 Aug 1999 22:34:53 +0000 (22:34 +0000)]
Add a VACUUM command in hopes of making pg_upgrade usable again
in MVCC environment. I do not trust this until Vadim says it's OK...
Michael Meskes [Mon, 2 Aug 1999 04:59:18 +0000 (04:59 +0000)]
*** empty log message ***
Tom Lane [Mon, 2 Aug 1999 02:05:41 +0000 (02:05 +0000)]
Further selectivity-estimation work. Speed up eqsel()
(it should just call the given operator, not look up an = operator).
Fix intltsel() so that all numeric data types are converted to double
before trying to estimate where the given comparison value is in the
known range of column values. intltsel() still needs work, or replacement,
for non-numeric data types ... but for nonintegral numeric types it
should now be delivering reasonable estimates.
Marc G. Fournier [Mon, 2 Aug 1999 01:42:57 +0000 (01:42 +0000)]
back out last...
Marc G. Fournier [Mon, 2 Aug 1999 01:40:30 +0000 (01:40 +0000)]
testing somethign...
Bruce Momjian [Sun, 1 Aug 1999 16:36:57 +0000 (16:36 +0000)]
autoconf
Bruce Momjian [Sun, 1 Aug 1999 16:30:05 +0000 (16:30 +0000)]
I didn't see any further discussion so here is, I hope, a clean fix to
configure.in to determine if a system is ELF or not. Note that some
of the tests earlier may be redundant but I took the safest route.
D'Arcy J.M. Cain
Bruce Momjian [Sun, 1 Aug 1999 14:48:50 +0000 (14:48 +0000)]
Update NT readme.
Tom Lane [Sun, 1 Aug 1999 04:54:25 +0000 (04:54 +0000)]
First step in fixing selectivity-estimation code. eqsel and
neqsel now behave as per my suggestions in pghackers a few days ago.
selectivity for < > <= >= should work OK for integral types as well, but
still need work for nonintegral types. Since these routines have never
actually executed before :-(, this may result in some significant changes
in the optimizer's choices of execution plans. Let me know if you see
any serious misbehavior.
CAUTION: THESE CHANGES REQUIRE INITDB. pg_statistic table has changed.
Tom Lane [Sun, 1 Aug 1999 04:41:32 +0000 (04:41 +0000)]
Add another ORDER BY to rules test; got tired of it 'failing'
every time I tweak the optimizer...
Tom Lane [Sat, 31 Jul 1999 22:06:44 +0000 (22:06 +0000)]
Re-enable pg_upgrade, after adding checks that the source
and target databases are of versions it knows about.
Tom Lane [Sat, 31 Jul 1999 22:05:48 +0000 (22:05 +0000)]
Add some more verbiage to man page's directions for pg_upgrade.
Tom Lane [Sat, 31 Jul 1999 19:07:25 +0000 (19:07 +0000)]
Add comments for attdisbursion field --- NO code change.
Tom Lane [Fri, 30 Jul 1999 22:34:19 +0000 (22:34 +0000)]
Update comments about clause selectivity estimation.
Bruce Momjian [Fri, 30 Jul 1999 18:09:49 +0000 (18:09 +0000)]
Make usecatupd disabled for normal users, and allow normal users to
update temp tables with this setting.
Tom Lane [Fri, 30 Jul 1999 04:07:25 +0000 (04:07 +0000)]
Further cleanups of indexqual processing: simplify control
logic in indxpath.c, avoid generation of redundant indexscan paths for the
same relation and index.
Marc G. Fournier [Fri, 30 Jul 1999 04:04:53 +0000 (04:04 +0000)]
There is one section that changed, concernign startup...the rest is just
changes for v6.5->v6.5.1, so relatively harmless
Marc G. Fournier [Fri, 30 Jul 1999 03:58:20 +0000 (03:58 +0000)]
v6.6's HISTORY file should reflect changes that went into all previous
releases, including v6.5.1 ...
Marc G. Fournier [Fri, 30 Jul 1999 03:45:57 +0000 (03:45 +0000)]
just testing a script...
Bruce Momjian [Fri, 30 Jul 1999 03:45:44 +0000 (03:45 +0000)]
Remove extra #endif
Marc G. Fournier [Fri, 30 Jul 1999 03:42:59 +0000 (03:42 +0000)]
Nothing changed, just testing cvslog ...
Bruce Momjian [Fri, 30 Jul 1999 00:59:50 +0000 (00:59 +0000)]
Fix description of psql flags -t and -q.
Tom Lane [Fri, 30 Jul 1999 00:56:17 +0000 (00:56 +0000)]
Fix coredump seen when doing mergejoin between indexed tables,
for example in the regression test database, try
select * from tenk1 t1, tenk1 t2 where t1.unique1 = t2.unique2;
6.5 has this same bug ...
Tom Lane [Fri, 30 Jul 1999 00:44:23 +0000 (00:44 +0000)]
Update comments for create_indexscan_node().
Bruce Momjian [Thu, 29 Jul 1999 15:39:05 +0000 (15:39 +0000)]
Remove comment after optimization flag.
Tom Lane [Thu, 29 Jul 1999 02:48:05 +0000 (02:48 +0000)]
Add support for Case exprs to fix_indxqual_references,
so that Case works in WHERE join clauses. Temporary patch --- this routine
is one of many that ought to be changed to use centralized expression-tree-
walking logic.
Tom Lane [Thu, 29 Jul 1999 02:45:36 +0000 (02:45 +0000)]
Add equal() funcs for Case nodes ... amazing we had not
detected this omission before. Miscellaneous other cleanups.
Bruce Momjian [Wed, 28 Jul 1999 17:52:31 +0000 (17:52 +0000)]
Update TODO list.
Bruce Momjian [Wed, 28 Jul 1999 17:51:41 +0000 (17:51 +0000)]
Update TODO list.
Tom Lane [Wed, 28 Jul 1999 17:39:38 +0000 (17:39 +0000)]
Allow a_expr not just AexprConst in the right-hand list of
IN and NOT IN operators. Rewrite grotty implementation of IN-list
parsing ... look Ma, no global variable ...
Bruce Momjian [Wed, 28 Jul 1999 00:09:49 +0000 (00:09 +0000)]
Update TODO list.
Tom Lane [Tue, 27 Jul 1999 06:23:12 +0000 (06:23 +0000)]
Correct bug in best_innerjoin(): it should check all the
rels that the inner path needs to join to, but it was only checking for
the first one. Failure could only have been observed with an OR-clause
that mentions 3 or more tables, and then only if the bogus path was
actually selected as cheapest ...
Tom Lane [Tue, 27 Jul 1999 03:53:54 +0000 (03:53 +0000)]
My recent optimizer changes caused a different plan to
be picked for one of the complex joins in rules test ... leading to
a different output ordering ...
Tom Lane [Tue, 27 Jul 1999 03:51:11 +0000 (03:51 +0000)]
First cut at doing LIKE/regex indexing optimization in
optimizer rather than parser. This has many advantages, such as not
getting fooled by chance uses of operator names ~ and ~~ (the operators
are identified by OID now), and not creating useless comparison operations
in contexts where the comparisons will not actually be used as indexquals.
The new code also recognizes exact-match LIKE and regex patterns, and
produces an = indexqual instead of >= and <=.
This change does NOT fix the problem with non-ASCII locales: the code
still doesn't know how to generate an upper bound indexqual for non-ASCII
collation order. But it's no worse than before, just the same deficiency
in a different place...
Also, dike out loc_restrictinfo fields in Plan nodes. These were doing
nothing useful in the absence of 'expensive functions' optimization,
and they took a considerable amount of processing to fill in.
Bruce Momjian [Mon, 26 Jul 1999 20:12:33 +0000 (20:12 +0000)]
Update TODO list.
Bruce Momjian [Mon, 26 Jul 1999 15:28:32 +0000 (15:28 +0000)]
Update TODO list.
Tom Lane [Sun, 25 Jul 1999 23:07:26 +0000 (23:07 +0000)]
Further work on planning of indexscans. Cleaned up interfaces
to index_selectivity so that it can be handed an indexqual clause list
rather than a bunch of assorted derivative data.
Tom Lane [Sun, 25 Jul 1999 17:53:27 +0000 (17:53 +0000)]
Remove 'restrictinfojoinid' field from RestrictInfo nodes.
The only place it was being used was as temporary storage in indxpath.c,
and the logic was wrong: the same restrictinfo node could get chosen to
carry the info for two different joins. Right fix is to return a second
list of unjoined-relids parallel to the list of clause groups.