I'm working on making some changes to the top level configure.in and m4
has now been running for 17 minutes on a 1.4G tbird. Am I missing
something or is this know to take forever?
For me, the autoconf run is instantaneous. Make sure you're using
Autoconf 2.13, and you don't have
I have added this to the TODO list:
* Add OR REPLACE clauses to non-FUNCTION object creation
I think there are clearly some other objects that need OR REPLACE. Not
sure which ones yet.
---
Dear all,
Would
Bruce Momjian [EMAIL PROTECTED] writes:
Our current CREATE OR REPLACE FUNCTION perserves the OID of the
function. Is there similar functionality you need where a simple
DROP (ignore the error), CREATE will not work?
If possible, it's nice to not have commands whose error codes you ignore.
On Tue, 23 Oct 2001, Bruce Momjian wrote:
Dear all,
Would it be possible to implement CREATE OR REPLACE VIEW / TRIGGER in
PostgreSQL 7.2?
Probably not, it's rather late in the cycle (isn't beta imminent?). Oh,
I'd vote for OR REPLACE as there's already an opt_or_replace
non-terminal in
Bruce Momjian writes:
I recommend tips when they are one line in length, have a high
probability of being accurate, and are common mistakes. Anything longer
and we should point to a specific section in the docs.
I would put when porting from MySQL into that category.
I would too
Not until we do the necessary legwork. I spent a good deal of time over
the past week making the various PL modules react to replacement of
pg_proc entries by CREATE OR REPLACE FUNCTION (cf. complaint from Peter
a week or so back). CREATE OR REPLACE VIEW implies updating cached
query
Bruce Momjian wrote:
Alternatively, could someone implement CREATE OR DROP VIEW / TRIGGER? These
features are needed for pgAdmin II (we could also provide a patch for
PhpPgAdmin). If this cannot be implemented in PostgreSQL, we will go for
pseudo-modification solutions (which is
Christopher Kings-Lynne [EMAIL PROTECTED] writes:
In the LOCK TABLE docs it documents the SELECT...FOR UPDATE as follows:
ROW SHARE MODE
Note: Automatically acquired by SELECT...FOR UPDATE. While it is a shared
lock, may be upgraded later to a ROW EXCLUSIVE lock.
Conflicts with EXCLUSIVE
I've got a bit of a problem. I added a fast SIGALRM handler in my
project to do various maintenance and this broke PQconnectStart().
Oct 23 21:56:36 james BlueList: connectDBStart() -- connect() failed:
Interrupted system call ^IIs the postmaster running (with -i) at
'archives.blue-labs.org'
+ What I really need is a binary *short* object type.
I have heard rumors of a legendary bytea type that might
help me, but it doesn't appear to be documented anywhere,
so I hesitate to use it.
It's real and it's not going away. It is pretty poorly documented
and doesn't have a
Tom Lane wrote:
Martin Weinberg [EMAIL PROTECTED] writes:
Yes, I understand locking the table, but empirically, two index
creations will not run simultaneously on the same table.
Hmm, on trying it you are right. The second index creation blocks here:
#6 0x1718e0 in XactLockTableWait
*very* slow, due to seq scan on
20 million entries, which is a test setup up to now)
Perennial first question: did you VACUUM ANALYZE?
Can there, or could there, be a notion of rule based optimization of
queries in PostgreSQL? The not using index problem is probably the
most
I'm working on an application where it is necessary to make copies of large
objects, and now I wonder if it is safe
to use this (symbolic, somewhat PHP like) code. Say I've a LOB with OID=1234
$oid = db_exec(select lo_create())
db_exec(delete from pg_largeobject where loid=$oid)
It seems Sybase has dropped the BETWEEN search condition. I thought
it was part of SQL92, has it been dropped from the spec since then or
wasn't it ever in there?
Vince.
--
==
Vince Vielhaber -- KA8CSHemail: [EMAIL
Hi all,
I was just looking for the code which checks for the memory available on
machine before writing the data.
Any related information will be appreciated.
Thanks,
KKG
_
Get your FREE download of MSN Explorer at
Hi all,
I need to perform a tree traversal on a big table (millions of rows).
To avoid recursive queries, one for each non-leaf node, this table has,
in addition to its 70 columns, a VARCHAR(3) column that is used
exclusively to sort the rows with the required order. The actual content
On Wed, 24 Oct 2001, Thomas Lockhart wrote:
It seems Sybase has dropped the BETWEEN search condition. I thought
it was part of SQL92, has it been dropped from the spec since then or
wasn't it ever in there?
It is documented in every SQL book I have and I see it in our SQL99
docs. Are
The minor featurette seems to have crept into current sources; it is
probably the cause of pg_dump being unable to reinstate disabled triggers.
Philip Warner| __---_
Albatross Consulting Pty. Ltd.
I'm working on an application where it is necessary to make copies of large
objects, and now I wonder if it is safe
to use this (symbolic, somewhat PHP like) code. Say I've a LOB with OID=1234
$oid = db_exec(select lo_create())
db_exec(delete from pg_largeobject where loid=$oid)
Joe Conway [EMAIL PROTECTED] writes:
I'll take a shot at improving the documentation for bytea. I'm hoping
documentation patches are accepted during beta though ;-)
Of course. The only limitation we place during beta is no new features
added. I plan to spend a good deal of time on the docs
HELP
left outer join intructionworking or not on POSTGRES 7
Zenon Karol
---(end of broadcast)---
TIP 2: you can get off all lists at once with the unregister command
(send unregister YourEmailAddressHere to [EMAIL PROTECTED])
bpalmer [EMAIL PROTECTED] writes:
I am running 2.13 (even on a clean checkout of 7.1.3) and the autoconf
takes forever. However, m4 is the process that's running forever, so I
have no doubs that the problem is there. What version do you use that
works?
GNU m4 ... the version I have here
mario [EMAIL PROTECTED] writes:
And another question regarding large objects, as I see the objects are
organized in units of 2048 bytes each. Can I somehow set this to a higher
value like 8k or 32k (I use 32k pages).
Then you've already got larger units, because the code is
#define LOBLKSIZE
Philip Warner [EMAIL PROTECTED] writes:
The minor featurette seems to have crept into current sources; it is
probably the cause of pg_dump being unable to reinstate disabled triggers.
Huh? There's never been a cast from int8 to int2. I checked 7.0 and
7.1, they both complain as well:
Zenon [EMAIL PROTECTED] writes:
left outer join intructionworking or not on POSTGRES 7
It works in 7.1 or later.
regards, tom lane
---(end of broadcast)---
TIP 3: if posting/reading through Usenet, please send an
I'd like to propose a new command, CREATE OPERATOR CLASS. Its purpose is
to create a named operator class, so that you can create new types of
index ops. Also, its inclusion would remove the section of the
documentation where we tell people how to manually manipulate the system
tables.
Since
Philip Warner [EMAIL PROTECTED] writes:
At 16:09 24/10/01 -0400, Tom Lane wrote:
Huh? There's never been a cast from int8 to int2. I checked 7.0 and
7.1, they both complain as well:
Is this a policy decision, or just a case where noone has had a chance to
do it?
Just a missing feature.
On Tue, 23 Oct 2001, Bill Studenmund wrote:
Here's the syntax I'd like to propose:
CREATE OPERATOR CLASS name [DEFAULT] FOR TYPE typename USING access
method WITH list of operators AND list of support functions
Hmmm.. Teach me to read the docs. :-) There's no way to set opckeytype. So
hwo
On Tue, Oct 23, 2001 at 08:43:32AM -0700, Bill Studenmund wrote:
And there's the fact that schemas were wanted for 7.2, and didn't happen.
Withouth external adgitation, will they happen for 7.3? Given the size of
the job, I understand why they didn't happen (the package changes so far
At 16:09 24/10/01 -0400, Tom Lane wrote:
Huh? There's never been a cast from int8 to int2. I checked 7.0 and
7.1, they both complain as well:
Is this a policy decision, or just a case where noone has had a chance to
do it?
Where exactly is pg_dump failing?
The problem in in the code to
On Wed, 24 Oct 2001, Tom Lane wrote:
Bill Studenmund [EMAIL PROTECTED] writes:
I'd like to propose a new command, CREATE OPERATOR CLASS.
Seems like a good idea.
operator spec is either an operator or an operator followed by the keyword
REPEATABLE. The presence of REPEATABLE indicates
Bill Studenmund [EMAIL PROTECTED] writes:
I'd like to propose a new command, CREATE OPERATOR CLASS.
Seems like a good idea.
operator spec is either an operator or an operator followed by the keyword
REPEATABLE. The presence of REPEATABLE indicates that amopreqcheck
should be set to true for
At 19:41 24/10/01 -0400, Tom Lane wrote:
We just had one earlier this week, so I suppose another wouldn't make
all that much difference. Comments?
My pref would be for the initdb; the current situation may break (other)
existing apps.
Bill Studenmund [EMAIL PROTECTED] writes:
[ revised proposal for CREATE OPERATOR CLASS syntax ]
I don't like the idea of writing a bunch of consecutive commas (and
having to count them correctly) for cases where we're inserting
noncontigous amopstrategy or amprocnum numbers. Perhaps the syntax
Arguile [EMAIL PROTECTED] writes:
If a table field is altered to add a default, the default value is
bypassed by pre-existing rules.
Yeah, this problem has been known for awhile (to me at least). The
difficulty is that default values are added to INSERTs by the parser,
which is before rule
Stephan Szabo [EMAIL PROTECTED] writes:
I think all we need to do to implement things correctly is to consider a
previous event only if both xmin and cmin of the old tuple match the
current xact command IDs, rather than considering it on the basis of
xmin alone.
Are there any things that
On Wed, 24 Oct 2001, Tom Lane wrote:
Stephan Szabo [EMAIL PROTECTED] writes:
I think all we need to do to implement things correctly is to consider a
previous event only if both xmin and cmin of the old tuple match the
current xact command IDs, rather than considering it on the basis of
Quick question - I couldn't find this in the docs:
What exactly is the advantage in using VIEWs? I get the impression that the
SELECT query it is based on is cached (ie. a cached query plan).
But, is this cached between db restarts, between connections, etc. Is it
cached upon the first use of
I get the impression that the
SELECT query it is based on is cached (ie. a cached query plan).
Nope. If there's something in the docs that makes you think so,
point out so I can fix it ;-)
Hmmm...I could have sworn that you mentioned in passing something about
cached query plans and
OK, I see my email got through to the list. Running pgindent now and
will commit changes.
--
Bruce Momjian| http://candle.pha.pa.us
[EMAIL PROTECTED] | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your
Tom Lane wrote:
Stephan Szabo [EMAIL PROTECTED] writes:
I think all we need to do to implement things correctly is to consider a
previous event only if both xmin and cmin of the old tuple match the
current xact command IDs, rather than considering it on the basis of
xmin alone.
Are
Just noticed this QT software:
http://www.globecom.net/tora/
It's a very lovely administrative tool for Oracle. I wonder if anyone would
be interested in porting it to Postgres?
Don't think many of the funky administrative functions can be acheived
remotely in Postgres yet tho?
Chris
I have run pgindent on the C files and run pgjindent on the jdbc files
as requested by the jdbc list. You can package up beta now. I will
update the HISTORY file tomorrow with recent changes.
--
Bruce Momjian| http://candle.pha.pa.us
[EMAIL PROTECTED]
Zeugswetter Andreas SB SD wrote:
*very* slow, due to seq scan on
20 million entries, which is a test setup up to now)
Perennial first question: did you VACUUM ANALYZE?
Can there, or could there, be a notion of rule based optimization of
queries in PostgreSQL? The not using
It seems Sybase has dropped the BETWEEN search condition. I thought
it was part of SQL92, has it been dropped from the spec since then or
wasn't it ever in there?
It is documented in every SQL book I have and I see it in our SQL99
docs. Are you *sure* Sybase dropped it? If so, then it
...
I'll take a shot at improving the documentation for bytea. I'm hoping
documentation patches are accepted during beta though ;-)
Always. At least up until a week or so before release, when we need to
firm up the docs and work on final cleanup etc. There are several
announcements leading up
Yes, I have seen this too today.
Is something amiss with the CVS server? I'm running an strace to watch
a cvs update and it's forbiddingly slow. It zooms along until it slams
into a brick wall for minutes, sometimes 10+ minutes, then it flys on.
David
Bill Studenmund [EMAIL PROTECTED] writes:
Do any of the access methods really support using non-binary operators?
Whether they do today is not the question. The issue is whether they
could --- and they certainly could.
Oh gross. I just looked at contrib/intarray, and it defines two entries
Tom Lane wrote:
"Hiroshi Inoue" [EMAIL PROTECTED] writes:
In the end, I changed DefineIndex() to not call IndexesAreActive().
I saw that. But is it a good solution? If someone has deactivated
indexes on a user table (ie turned off relhasindex), then creating a
new index would activate
On Wed, 24 Oct 2001, Tom Lane wrote:
The only reason we do this is to enforce the triggered data change
violation restriction of the spec. However, I think we've
misinterpreted the spec. The code prevents an RI referenced value from
being changed more than once in a transaction, but what
-Original Message-
From: Hiroshi Inoue
Tom Lane wrote:
Martin Weinberg [EMAIL PROTECTED] writes:
Yes, I understand locking the table, but empirically, two index
creations will not run simultaneously on the same table.
Hmm, on trying it you are right. The second index
On Tue, 23 Oct 2001, Bruce Momjian wrote:
If possible, it's nice to not have commands whose error codes you ignore.
That way if you see an error, you know you need to do something about it.
Folks, is this a valid reason for adding OR REPLACE to all CREATE object
commands?
Sounds good to
David Ford [EMAIL PROTECTED] writes:
I've got a bit of a problem. I added a fast SIGALRM handler in my project to
do various maintenance and this broke PQconnectStart().
Oct 23 21:56:36 james BlueList: connectDBStart() -- connect() failed:
Interrupted system call ^IIs the postmaster
Bill Studenmund writes:
Why? Operators are used differently than functions.
I don't think so. Operators are a syntacticaly convenience for functions.
That's what they always have been and that's what they should stay.
How does what you say disagree with what I said?
Operators
David Ford [EMAIL PROTECTED] writes:
I've got a bit of a problem. I added a fast SIGALRM handler in my
project to do various maintenance and this broke PQconnectStart().
It'd probably be reasonable to just retry the connect() call if it
fails with EINTR. If that works for you, send a
On Wed, 24 Oct 2001, Peter Eisentraut wrote:
Bill Studenmund writes:
So I am a naive programmer because I mention intent above?
No.
Sorry, that's the way it came across. As you've said that was not your
intent, please disregard my response; I was responding to something you
did not mean.
Lee Kindness writes:
After a simple './configure' on a stock Solaris 2.6 box the
compilation of interfaces/ecpg/lib/execute.c fails due to the macro
definition of 'gettext' to ''. This macro is invoked on the prototype
of gettext() in libintl.h (included via locale.h).
This should be fixed
mlw writes:
The not using index was very frustrating to understand. The stock answer,
did you vacuum? just isn't enough. There has to be some explanation (in the
FAQ or something) about the indexed key distribution in your data.
Most not using index questions seem to be related to a
Bill Studenmund writes:
Our current CREATE OR REPLACE FUNCTION perserves the OID of the
function. Is there similar functionality you need where a simple
DROP (ignore the error), CREATE will not work?
If possible, it's nice to not have commands whose error codes you ignore.
That way if
I have been looking at the way that deferred triggers slow down when the
same row is updated multiple times within a transaction. The problem
appears to be entirely due to calling deferredTriggerGetPreviousEvent()
to find the trigger list entry for the previous update of the row: we
do a linear
mlw [EMAIL PROTECTED] writes:
... Postgres' statistics are pretty poor too, a relative few very
populous entries in a table will make it virtually impossible for the
cost based optimizer (CBO) to use an index.
Have you looked at development sources lately?
regards,
I have been asked to run pgindent in preparation for beta starting
tomorrow. In this run, I will also reformat the jdbc files as agreed to
by the jdbc list. I don't have much time to wait before starting the
pgindent run. I hope people don't have outstanding patches sitting
around.
--
Hiroshi Inoue [EMAIL PROTECTED] writes:
In the end, I changed DefineIndex() to not call IndexesAreActive().
I saw that. But is it a good solution? If someone has deactivated
indexes on a user table (ie turned off relhasindex), then creating a
new index would activate them again, which would
Jason Orendorff writes:
Hi. I was surprised to discover today that postgres's
character types don't support zero bytes. That is,
Postgres isn't 8-bit clean. Why is that?
PostgreSQL is 8-bit clean. The character types don't support zero bytes
because the character types store characters,
Um, Vadim? Still of the opinion that elog(STOP) is a good
idea here? That's two people now for whom that decision has
turned localized corruption into complete database failure.
I don't think it's a good tradeoff.
One is able to use pg_resetxlog so I don't see point in
removing
65 matches
Mail list logo