Обсуждение: 6.3 patches
Here are some of the fixes in 6.3 if people want to test them.
The only problem I know of is that:
update test set x = max(test2.y)
only updates one row each time it is run. Any ideas on a fix?
---------------------------------------------------------------------------
RELIABILITY
-----------
* Overhaul mdmgr/smgr to fix double unlinking and double opens, cleanup
* Overhaul bufmgr/lockmgr/transaction manager
* -Fix CLUSTER
* Remove EXTEND?
* -Aggregates on VIEW always returns zero (maybe because there is no oid for views?)
* CREATE VIEW requires super-user priviledge
* can lo_export()/lo_import() read/write anywhere, causing a security problem?
* Tables that start with xinv confused to be large objects
* Two and three dimmensional arrays display improperly, missing {}
* -Add GROUP BY to INSERT INTO table SELECT * FROM table2
* lo_unlink() crashes server
* Prevent auto-table reference, like SELECT table.col WHERE col = 3 (?)
* -Remove un-needed malloc() calls and replace with palloc().
* SELECT * FROM table WHERE int4_column = '1' fails
* SELECT a[1] FROM test fails, it needs test.a[1]
* -SELECT COUNT(*) FROM TAB1, TAB2 fails
* -SELECT SUM(2+2) FROM table dumps core
* UPDATE table SET table.value = 3 fails
* -UPDATE key_table SET keyval=count(reftab.num) fails
* -INSERT INTO ... SELECT DISTINCT ... does not accept DISTINCT
* -INSERT INTO table SELECT id, count(*) FROM table2 GROUP BY id generates error
* Make pg_dump preserve inheritance column order, do non-inherits first
* User who can create databases can modify pg_database table
--
Bruce Momjian
maillist@candle.pha.pa.us
Some more items from my list which have been added since v6.2.1:
SQL92 binary and hex input and string type coersion
allow casting of non-constants using both SQL92 and Postgres syntax
parser support for PRIMARY, FOREIGN KEY
backend support for PRIMARY KEY, UNIQUE (create index)
more parser support for DEFAULT, CHECK clauses
add 'doy' as argument to datetime_part()
functions datetime_time(), time(datetime)
functions int4_datetime(), int4_timespan()
add unixdate package to contrib area
constant CURRENT_USER as GetPgUserName()
support SQL3 syntax TRUE, FALSE
implement IS TRUE, IS FALSE, IS NOT TRUE, IS NOT FALSE
provide timezone support in libpq and backend
support SQL92 "delimited identifiers" (libpq, scan.l, psql)
PGTZ environment variable for frontend
fix session initialization from front-end environment variables
use PGDATESTYLE to initialize backend at startup
add GERMAN date style
allow alternate locations for databases
add hash functions for datetime, timespan
fix hash function declarations for float8 and int4
convert text to/from int4, int2
Also, we've added UNIONs (didn't see it on Bruce's list).
- Tom
> > Some more items from my list which have been added since v6.2.1: > > SQL92 binary and hex input and string type coersion > allow casting of non-constants using both SQL92 and Postgres syntax > parser support for PRIMARY, FOREIGN KEY > backend support for PRIMARY KEY, UNIQUE (create index) > more parser support for DEFAULT, CHECK clauses > add 'doy' as argument to datetime_part() > functions datetime_time(), time(datetime) > functions int4_datetime(), int4_timespan() > add unixdate package to contrib area > constant CURRENT_USER as GetPgUserName() > support SQL3 syntax TRUE, FALSE > implement IS TRUE, IS FALSE, IS NOT TRUE, IS NOT FALSE > provide timezone support in libpq and backend > support SQL92 "delimited identifiers" (libpq, scan.l, psql) > PGTZ environment variable for frontend > fix session initialization from front-end environment variables > use PGDATESTYLE to initialize backend at startup > add GERMAN date style > allow alternate locations for databases > add hash functions for datetime, timespan > fix hash function declarations for float8 and int4 > convert text to/from int4, int2 > > Also, we've added UNIONs (didn't see it on Bruce's list). I was just highlighting fixes to things currently on the TODO list under reliability. -- Bruce Momjian maillist@candle.pha.pa.us