Updated: http://www.sigaev.ru/misc/fast_insert_gin-0.9.gz
need more review of fast_insert yet? It looked like a number of people
commented on it already.
I still havn't clearness of acceptability for suggested aminsertcleanup
calling.
--
Teodor S
pport/synchronize with multicolumn GIN - both patches
touch the same pieces of code, and I didn't make a single patch to simplify review.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
-
I looked this over and it looks good in general.
May I think that patch passed review and commit it?
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list
index is binary compatible with current index :)
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To make changes to your subs
*1.885* ms4.994 ms
Index: ~340 s~200 s
Insert: 72 s/166 s/1
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@pos
Sync with current CVS HEAD and post in hackers- too because patches- close to
the closing.
http://www.sigaev.ru/misc/fast_insert_gin-0.7.gz
http://www.sigaev.ru/misc/multicolumn_gin-0.3.gz
--
Teodor Sigaev E-mail: [EMAIL PROTECTED
nsert into GIN' patch with CVS.
Objections?
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To make
ith a fake Relation struct, and
findParents will call ReadBuffer()...
GiST's function gistContinueInsert() is working by similar way.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent vi
is much-much more important (or crucial) than
insertion time. Or table stores read-only values.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (
structure then it's time to call insert cleanup.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To make
ay take several
seconds in unpredictable moment.
the buffer. To keep that overhead out of the main codepath, we could
make autovacuum to flush the buffers periodically.
Do you mean that GIN sends a "smoke signal" to the autovacuum launcher process
to ask about v
ethod was introduced.
- added documentation
Suppose, patch is ready to review/commit...
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-pa
(fast insertion) feature?
If so, what is a default value? It's not clear to me.
Note: These patches are mutually exclusive because they touch the same pieces
of code and I'm too lazy to manage several depending patches. I don't see any
problem to join patc
There seems to be something broken here: it's acting like prefix search
is on all the time, eg
I'm in sackcloth and ashes...
Fixed and extended regression tests.
http://www.sigaev.ru/misc/tsearch_prefix-0.9.gz
--
Teodor Sigaev E-mail: [EMAIL
http://www.sigaev.ru/misc/wildspeed-0.12.tgz
Changes:
- Sync with CVS HEAD
- add third option (StrategyNumber) for comparePartialFn.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW:
http://www.sigaev.ru/
--
Se
27;ll add
this option.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To make changes to your subscription:
http://www
http://www.sigaev.ru/misc/partial_match_gin-0.9.gz
Sync with CVS.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To
d-0.11.tgz
Sync with CVS changes and partial match GIN patch. Teach opclass to correct use
of recheck feature.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches ma
g to cost light optimization...
This is still WIP because I haven't touched any contrib code, but
as far as the main backend goes I think it's ready to apply.
Patch to all contrib modules:
http://www.sigaev.ru/misc/contrib.patch.gz
--
Teodor Sigaev
could satisfy '%foo%' using a regular and a reverse B-tree index,
and a bitmap AND. Which is interestingly similar to the way you proposed
to use a TIDBitmap within GIN.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
is impossible for btree. But I don't like a
hack around LIKE support in BTree. This support uses outflank ways missing
regular one.
I'm thinking about add new strategy to Btree and allow directly support of
prefix LIKE search. And BTree will scan index while compare method with option
indexqual_opclause().
NOTICE 2: it seems to me, that similar technique could be implemented for
ordinary BTree to eliminate hack around LIKE support.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.
suspiciously long for human input.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing list (pgsql-patches@postgresql.org)
To make changes to your subscription:
http
account Clinton'
For another dictionary ( dictionary of number, snowball ) that option is a
meaningless.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent via pgsql-patches mailing lis
ALTER TEXT SEARCH DICTIONARY foo (...) WITH ( filtering=on|off,
store_original=on|off );
Or per token's type/dictionary pair.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
--
Sent vi
_type.h) field. It
contains number of bytes between first lexeme in tsvector and needed lexeme.
So, limitation is total length of lexemes plus theirs positional information.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
notebook: Core2 Duo 1.8MHz, 2Gb with default postgres.conf
8.4 without patch:
Time: 10883,368 ms
8.4 with patch (db was reinited)
Time: 9654,266 ms
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW:
Magnus, could you provide dump of tsvector column of archive search?
I'll make a test with and without Tom's patch.
I have not done any performance testing of these changes --- does
anyone have specific test scenarios they'd like to try?
--
Teodor Sigaev
line and rank reads a lot of tsvectors. It seems to me that ranking test
will be very clear: rank function reads whole tsvector and returns small amount
of data (just a number).
Another testing focus may be a lossy indexes, like a index over polygons.
--
Teodor Sigaev
--
---(end of broadcast)---
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
--
Teodor Sigaev
This change introduced a compiler warning. Here is a fix for it.
Remove QueryOperand->istrue flag, it was used only in cover ranking
(ts_rank_cd). Use palloc'ed array in ranking instead of flag.
Thank you, committed
--
Teodor Sigaev E-mail: [EMAIL P
memory,
but is more readable in source code. The POSDATAPTR and POSDATALEN
macros are still used, though it would now be more readable to access
the fields in WordEntryPosVector directly.
Did you check it on 64-bit boxes with strict alignment? I remember that was a
headache for me.
--
Teodor
I am getting confused with the patches and version I have lying around
here... I think I'll have to wait for review of the patches I've posted
this far before I continue hacking.
Sorry for delay - I was busy by another. All your patches are committed with
very small changes.
--
Teo
BTW, Teodor, are you intending to review/apply Heikki's tsearch fixes,
or do you want someone else to do it?
I'll do it.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www
Heikki Linnakangas wrote:
Teodor Sigaev wrote:
Ok. Probably easiest to do that by changing the palloc to palloc0 in
parse_tsquery.
and change sizeof to sizeof(QueryItem)
Do you mean the sizeofs in the memcpys in parse_tsquery? You can't
Oops, I meant pallocs in push* function. pallo
Ok. Probably easiest to do that by changing the palloc to palloc0 in
parse_tsquery.
and change sizeof to sizeof(QueryItem)
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru
ns I found.
Some of them might have a natural limit so that you can't force
arbitrarily deep recursions, but check_stack_depth() is cheap enough
that seems best to just stick it into anything that might be a problem.
Agreed.
--
Teodor Sigaev E-mail: [
objections, QTNode and NODE are tree-like structures, but
TSQuery is a postfix notation for storage in plain memory. NODE is used only
cleanup stop-word placeholders, so it's a binary tree while QTNode represents
t-ary tree (with any number of children).
Thank you for your interesting in
Tom Lane wrote:
Teodor Sigaev <[EMAIL PROTECTED]> writes:
http://www.sigaev.ru/misc/tsearch_core-0.58.gz
What is src/backend/utils/tsearch/dict_ispell/parse.h ?
oops - that is unused file after some experiments. Will be removed from patch.
Well, I know what it *is*: it's b
arch-config=CFG switch.
6) pg_dump, psql are changed accordingly.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
---(end of broadcast)---
TI
ScanKey->sk_strategy = InvalidStrategy and btree code transforms it to
BTEqualStrategyNumber in _bt_preprocess_keys(), btcostestimate knows that fact too.
GiST doesn't need to transform strategy - it looks at SK_INDEXFINDNULL only.
--
Teodor Sigaev
nd words
( German, Norwegian ). So, please, test it - we don't know that languages at
all.
2 added recent fixes of contrib/tsearch2
3 fix usage of fopen/fclose
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
& (a->lower <= b->upper))
||
((b->upper >= a->upper) && (b->lower <= a->upper))
) ? true : false;
}
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http:
I don't very like GiST changes: gistgetbitmap will lock/unlock page for every
tuple. It seems to me taht is better to change gistnext function to use some
sort callback, for example. Or have static array for tids in gistgetbitmap...
--
Teodor Sigaev E
I think it can be interesting for other flavours of GIN usage. Is
there a way to add the number of entries of the considered indexed
item to the consistent prototype without adding too much overhead and
complexity?
We are thinking about adding extra value, but it's still only thinking.
---
matter for index scan.
The attached patch adds a RECHECK too. It seems to work correctly but
the RECHECK COND costs a lot of time :/.
:(
How long is average length of strings in table?
--
Teodor Sigaev E-mail: [EMAIL
d mark '%'
operator in CREATE OPERATOR CLASS with RECHECK option.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
---(end of broadcast)---
TIP 2: Don't 'kill -9' the postmaster
> representations, both redundant with the textual description. I don't
Docs patch is in SGML table representation, text view is a demonstration in
mail.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
W
| X |
+---+---+---+---+---+---+---+---+
Upper row and left column contain abbreviation of lock modes by the first
characters( AS - ACCESS SHARE )
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.siga
Nice, thanks a lot.
Tom Lane wrote:
Teodor Sigaev <[EMAIL PROTECTED]> writes:
Just a freshing for clean applying..
http://www.sigaev.ru/misc/user_defined_typmod-0.11.gz
Applied with some revisions, and pg_dump support and regression tests
added.
regards, to
This is not responding to my concern. What you presented was an
> Sorry, I see your point now.
Is that test enough? Or I should make more?
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev
Just a freshing for clean applying..
http://www.sigaev.ru/misc/user_defined_typmod-0.11.gz
Is any objections to commit?
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru
0.9 doesn't apply cleanly after Peter's changes, so, new version
http://www.sigaev.ru/misc/user_defined_typmod-0.10.gz
Teodor Sigaev wrote:
>> Perhaps an array of int4 would be better? How much
Done
http://www.sigaev.ru/misc/user_defined_typmod-0.9.gz
The patch needs mor
>> Perhaps an array of int4 would be better? How much
Done
http://www.sigaev.ru/misc/user_defined_typmod-0.9.gz
The patch needs more cleanup before applying, too, eg make comments
match code, get rid of unused keywords added to gram.y.
Cleaned.
--
Teodor
0) or
(f1>8 and f1<9)order by f1, f2 limit 10;
NOTICE: Elapsed time 0.000962 sec
3) 4 clauses
# select * from foo where (f1>4 and f1<5) or (f1>6 and f1<7) or
(f1>8 and f1<9) or (f1>95000 and f1<96000) order by f1, f2 limit 10;
rows=10
loops=1)
-> Result (cost=0.00..6902.27 rows=3928 width=8) (actual
time=0.111..0.197 rows=10 loops=1)
-> Append (cost=0.00..6902.27 rows=3928 width=8) (actual
time=0.102..0.156 rows=10 loops=1)
-> Index Scan using idx on foo (cost=0.00
Is there a guarantee that hash value of some not-null keys doesn't equal to
special hash code?
the patch to do IS NULL only. But if we are going areto support both,
we probably have to have two pg_am flags not one.
GiST isn't effective with single NOT NULL condition ... So,
sult is still going to have to pack into 32 bits.
As I remember, I tried to use some thing else but, I've got a lot conflicts
with
AexprConst:
func_name '(' expr_list ')' Sconst
The patch needs more cleanup before applying, too, eg make comments
match code, g
rantee that target lists are the same. Sorry, but I
didn't find clearer solution...
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
---(end of broadcast)---
TIP 2: Don't 'kill -9' the postmaster
or-last into the planner's model of sort order,
Agree, but I tried to keep patches independent as possible...
If we will have agreement about ways to resolve, I'll will time to work
further in foreseeable future.
--
Teodor Sigaev
t; 60100 )
ORDER BY a
LIMIT 20;
Limit (cost=0.00..4.14 rows=20 width=4) (actual time=0.168..1.001 rows=20
loops=1)
-> Index Scan using aidx on tst (cost=0.00..2344.85 rows=11338 width=4)
(actual time=0.162..0.935 rows=20 loops=1)
Index Cond: ((a > 5) AND (a < 6100
Patch isn't full, simple test (values are took from regression.diffs):
and try dump table and restore:
ERROR: syntax error
CONTEXT: COPY tt, line 5, column tq: "'1 ''2'"
Attached cumulative patch fixes problem, but I have some doubts, is it
3'' 4 ', NULL);
# insert into tt values ( NULL, E'''1 \\''2''');
# insert into tt values ( NULL, E'''1 \\''2''');
# insert into tt values ( NULL, E'1&(''2''&a
ubdir = contrib/tsearch2/wordparser
top_builddir = ../../..
include $(top_builddir)/src/Makefile.global
include $(top_srcdir)/contrib/contrib-global.mk
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http:
nd our makefile structure.
Please consider applying this to 8.1 (or just let me know and I'll do it
for you).
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
65 matches
Mail list logo
