On Sun, 2006-01-15 at 18:03 -0500, Tom Lane wrote:
There's one nontrivial decision still to make about how to implement
proper per-spec row-comparison operations, namely: how a row
comparison ought to be represented in the index access method API.
I'm not sure I understand why. Surely a row
On Mon, 2006-01-16 at 00:07 -0500, Rod Taylor wrote:
A couple of days ago I found myself wanting to aggregate 3 Billion
tuples down to 100 Million tuples based on an integer key with six
integer values -- six sum()'s.
PostgreSQL ran out of memory with its Hash Aggregator and doing an old
On Sun, 2006-01-15 at 11:14 -0500, korry wrote:
I've noticed a variety of coding styles in the PostgreSQL source code. In
particular, I see a mix of naming conventions. Some variables use camelCase
(or CamelCase), others use under_score_style.
I just follow the style of the module I'm
I've created a browsable source tree documentation, it's done with the
doxygen tool.
http://www.mcknight.de/pgsql-doxygen/cvshead/html/
There was a discussion about this some time ago, Jonathan Gardner proposed
it here:
http://archives.postgresql.org/pgsql-hackers/2004-03/msg00748.php
quite a
I wish I've had this when I started working with PostgreSQL. This looks really good. Very
useful indeed, even without the comments. What kind of changes are needed in order to get
the comments in?
Regards,
Thomas Hallgren
Joachim Wieland wrote:
I've created a browsable source tree
Try following the link (the Doxygen icon) - it has both a tutorial and
extensive doc.
Regards,
Kim Bisgaard
Thomas Hallgren wrote:
I wish I've had this when I started working with PostgreSQL. This
looks really good. Very useful indeed, even without the comments. What
kind of changes are
Thomas Hallgren said:
I wish I've had this when I started working with PostgreSQL. This looks
really good. Very useful indeed, even without the comments. What kind
of changes are needed in order to get the comments in?
I too have done this. But retrofitting Doxygen style comments to the
Title: [PATCH] Better way to check for getaddrinfo function.
Just thought that the following patch might improve checking for getaddrinfo function (in configure.in)
I was forced to write 'coz getaddrinfo went unnoticed in Tru64 Unix.
(displaying attached patch)
$ diff -r configure.in
On Mon, 2006-01-16 at 08:32 +, Simon Riggs wrote:
On Mon, 2006-01-16 at 00:07 -0500, Rod Taylor wrote:
A couple of days ago I found myself wanting to aggregate 3 Billion
tuples down to 100 Million tuples based on an integer key with six
integer values -- six sum()'s.
PostgreSQL ran
On Sun, Jan 15, 2006 at 06:03:12PM -0500, Tom Lane wrote:
There's one nontrivial decision still to make about how to implement
proper per-spec row-comparison operations, namely: how a row
comparison ought to be represented in the index access method API.
The current representation of index
Simon Riggs [EMAIL PROTECTED] writes:
On Sun, 2006-01-15 at 18:03 -0500, Tom Lane wrote:
There's one nontrivial decision still to make about how to implement
proper per-spec row-comparison operations, namely: how a row
comparison ought to be represented in the index access method API.
I'm
http://www.postgresql.org/docs/8.1/static/indexes-types.html
says:
The optimizer can also use a B-tree index for queries involving the
pattern matching operators LIKE, ILIKE, ~, and ~*, if the pattern is a
constant and is anchored to the beginning of the string - for example,
col LIKE 'foo%' or
Martijn van Oosterhout kleptog@svana.org writes:
ISTM that row-wise comparisons, as far as indexes are concerned are
actually simpler than normal scan-keys. For example, if you have the
condition (a,b) = (5,1) then once the index has found that point,
every subsequent entry in the index
Simon Riggs [EMAIL PROTECTED] writes:
On Mon, 2006-01-16 at 00:07 -0500, Rod Taylor wrote:
A couple of days ago I found myself wanting to aggregate 3 Billion
tuples down to 100 Million tuples based on an integer key with six
integer values -- six sum()'s.
There is already hash table overflow
R, Rajesh (STSD) [EMAIL PROTECTED] writes:
Just thought that the following patch might improve checking for
getaddrinfo function (in configure.in)
Since AC_TRY_RUN tests cannot work in cross-compilation scenarios,
you need an *extremely* good reason to put one in. I thought this
might improve
Magnus Hagander [EMAIL PROTECTED] writes:
http://www.postgresql.org/docs/8.1/static/indexes-types.html
says:
The optimizer can also use a B-tree index for queries involving the
pattern matching operators LIKE, ILIKE, ~, and ~*, if the pattern is a
constant and is anchored to the beginning of
On Sat, Jan 14, 2006 at 04:13:56PM -0500, Qingqing Zhou wrote:
Qingqing Zhou [EMAIL PROTECTED] wrote
I wonder if we should really implement file-system-cache-warmup strategy
which we have discussed before. There are two natural good places to do
this:
(1) sequentail scan
(2)
On Fri, Jan 13, 2006 at 11:37:38PM -0500, Tom Lane wrote:
Josh Berkus josh@agliodbs.com writes:
It's also worth mentioning that for datatypes that only have an =
operator the performance of compute_minimal_stats is O(N^2) when values
are unique, so increasing sample size is a very bad idea
On Mon, 2006-01-16 at 07:57 -0600, Andrew Dunstan wrote:
I too have done this. But retrofitting Doxygen style comments to the
PostgreSQL source code would be a big undertaking. Maintaining it, which
would be another task for reviewers/committers, would also be a pain unless
there were some
Neil Conway [EMAIL PROTECTED] writes:
I don't think it would be all that painful. There would be no need to
convert the entire source tree to use proper Doxygen-style comments in
one fell swoop: individual files and modules can be converted whenever
anyone gets the inclination to do so. I
On Sat, Jan 14, 2006 at 07:28:21PM +0900, Michael Glaesemann wrote:
On Jan 13, 2006, at 21:42 , Leandro Guimar?es Faria Corcete DUTRA wrote:
If you still declare the natural key(s) as UNIQUEs, you have just made
performance worse. Now there are two keys to be checked on UPDATEs
and
On Mon, 2006-01-16 at 09:42 -0500, Rod Taylor wrote:
On Mon, 2006-01-16 at 08:32 +, Simon Riggs wrote:
On Mon, 2006-01-16 at 00:07 -0500, Rod Taylor wrote:
A question: Are the rows in your 3 B row table clumped together based
upon the 100M row key? (or *mostly* so) We might also be
On Mon, 2006-01-16 at 12:36 -0500, Tom Lane wrote:
Simon Riggs [EMAIL PROTECTED] writes:
On Mon, 2006-01-16 at 00:07 -0500, Rod Taylor wrote:
A couple of days ago I found myself wanting to aggregate 3 Billion
tuples down to 100 Million tuples based on an integer key with six
integer
[EMAIL PROTECTED] wrote:
NT4 is officially dead, IMHO no need for PostgreSQL to officially support it,
let's leave place for companies offering commercial postgresql versions to
work on it if they have enough customer requests.
BTW Win 2000 is more or less 6 years old now ...
Agreed. If they
Bruce Momjian wrote:
[EMAIL PROTECTED] wrote:
NT4 is officially dead, IMHO no need for PostgreSQL to officially support it,
let's leave place for companies offering commercial postgresql versions to
work on it if they have enough customer requests.
BTW Win 2000 is more or less 6 years old now
NT4 is officially dead, IMHO no need for PostgreSQL to officially
support it, let's leave place for companies offering commercial
postgresql versions to work on it if they have enough
customer requests.
BTW Win 2000 is more or less 6 years old now ...
I believe Microsoft has an
Simon Riggs [EMAIL PROTECTED] writes:
For HJ we write each outer tuple to its own file-per-batch in the order
they arrive. Reading them back in preserves the original ordering. So
yes, caution required, but I see no difficulty, just reworking the HJ
code (nodeHashjoin and nodeHash). What else
On Mon, 2006-01-16 at 12:26 -0600, Jim C. Nasby wrote:
On Fri, Jan 13, 2006 at 11:37:38PM -0500, Tom Lane wrote:
Josh Berkus josh@agliodbs.com writes:
It's also worth mentioning that for datatypes that only have an =
operator the performance of compute_minimal_stats is O(N^2) when values
On Fri, 13 Jan 2006 19:18:29 +, Simon Riggs
[EMAIL PROTECTED] wrote:
I enclose a patch for checking out block sampling.
Can't comment on the merits of block sampling and your implementation
thereof. Just some nitpicking:
|! * Row Sampling: As of May 2004, we use the Vitter algorithm to
Simon Riggs [EMAIL PROTECTED] writes:
Tom has not spoken against checking for UNIQUE constraints: he is just
pointing out that there never could be a constraint in the case I was
identifying.
More generally, arguing for or against any system-wide change on the
basis of performance of
I'm considering getting rid of the BTItem/BTItemData and
HashItem/HashItemData struct definitions and just referencing
IndexTuple(Data) directly in the btree and hash AMs. It appears that
at one time in the forgotten past, there was some access-method-specific
data in index entries in addition to
On Mon, Jan 16, 2006 at 03:52:01PM -0500, Tom Lane wrote:
I'm considering getting rid of the BTItem/BTItemData and
HashItem/HashItemData struct definitions and just referencing
IndexTuple(Data) directly in the btree and hash AMs. It appears
that at one time in the forgotten past, there was
David Fetter [EMAIL PROTECTED] writes:
On Mon, Jan 16, 2006 at 03:52:01PM -0500, Tom Lane wrote:
Does anyone see a reason to keep this layer of struct definitions?
If you cut it out, what will the heap and index access methods
needed for SQL/MED use?
What's that have to do with this?
>From what I've seen, I don't think we need to keep them around.On 1/16/06, Tom Lane [EMAIL PROTECTED] wrote:
I'm considering getting rid of the BTItem/BTItemData andHashItem/HashItemData struct definitions and just referencing
IndexTuple(Data) directly in the btree and hash AMs.It appears thatat
On Mon, Jan 16, 2006 at 04:02:07PM -0500, Tom Lane wrote:
David Fetter [EMAIL PROTECTED] writes:
On Mon, Jan 16, 2006 at 03:52:01PM -0500, Tom Lane wrote:
Does anyone see a reason to keep this layer of struct
definitions?
If you cut it out, what will the heap and index access methods
not sure if this is the right place to post...
I am using postgres 8.1. In indxpath.c, it says Note: if Postgres tried to optimize queries by forming equivalenceclasses over equi-joined attributes (i.e., if it recognized that a
qualification such as where a.b=c.d and a.b=5 could make use ofan
David Fetter [EMAIL PROTECTED] writes:
On Mon, Jan 16, 2006 at 04:02:07PM -0500, Tom Lane wrote:
David Fetter [EMAIL PROTECTED] writes:
If you cut it out, what will the heap and index access methods
needed for SQL/MED use?
What's that have to do with this?
I'm sure you'll correct me if
uwcssa [EMAIL PROTECTED] writes:
I am using postgres 8.1. In indxpath.c, it says Note: if Postgres tried
to optimize queries by forming equivalence
classes over equi-joined attributes (i.e., if it recognized that
aqualification such as where
a.b=3Dc.d and a.b=3D5 could make use of
an
On Mon, Jan 16, 2006 at 12:07:44PM -0500, Tom Lane wrote:
Since you didn't understand what I was saying, I suspect that plan A is
too confusing ...
Umm, yeah. Now you've explained it I think it should be excluded on the
basis that it'll be a source of bugs. For all the places that matter a
On Mon, Jan 16, 2006 at 04:21:50PM -0500, Tom Lane wrote:
David Fetter [EMAIL PROTECTED] writes:
On Mon, Jan 16, 2006 at 04:02:07PM -0500, Tom Lane wrote:
David Fetter [EMAIL PROTECTED] writes:
If you cut it out, what will the heap and index access
methods needed for SQL/MED use?
On Jan 17, 2006, at 3:51 , Tom Lane wrote:
A quick look through the doxygen manual doesn't make it sound too
invasive, but I am worried about how well it will coexist with
pgindent.
It seems both tools think they can dictate the meaning of the
characters
immediately after /* of a comment
Martijn van Oosterhout kleptog@svana.org writes:
On Mon, Jan 16, 2006 at 12:07:44PM -0500, Tom Lane wrote:
Since you didn't understand what I was saying, I suspect that plan A is
too confusing ...
Umm, yeah. Now you've explained it I think it should be excluded on the
basis that it'll be a
On Tue, Jan 17, 2006 at 06:51:15AM +0900, Michael Glaesemann wrote:
I haven't looked at it yet, but might there not be a way to have a
preprocessing step where the current comment format is converted to
something doxygen-friendly? pg_indent2doxygen or something? Then the
current comment style
Wow, looks great. Is that URL stable? Can we link to it from the
PostgreSQL developers page?
http://www.postgresql.org/developer/coding
---
Joachim Wieland wrote:
I've created a browsable source tree
the only question I have ... is there any way of improving that right
index? Love the 'detail pages', but somehow making the right index more
'tree like' might make navigation a bit easier ...
On Mon, 16 Jan 2006, Bruce Momjian wrote:
Wow, looks great. Is that URL stable? Can we link
Fine. The rest documentation says:For now, the test only uses restriction clauses (those in restrictinfo_list). --Nels, Dec '92, however, I understand it as being overridden by the
followup, which is:XXX as of 7.1, equivalence class info *is* available. Considerimproving this code as foreseen by
On Jan 17, 2006, at 8:40 , Marc G. Fournier wrote:
the only question I have ... is there any way of improving that
right index? Love the 'detail pages', but somehow making the right
index more 'tree like' might make navigation a bit easier ...
Along those lines, I wonder if a CSS
On Mon, 2006-01-16 at 21:24 +0100, Manfred Koizar wrote:
On Fri, 13 Jan 2006 19:18:29 +, Simon Riggs
[EMAIL PROTECTED] wrote:
I enclose a patch for checking out block sampling.
Can't comment on the merits of block sampling and your implementation
thereof.
But your thoughts are
This was my plan all along, just been waiting for someone to make it work with
the postgresql code and then send instructions to the postgresql web team on
how to set it up.
Robert Treat
On Monday 16 January 2006 18:32, Bruce Momjian wrote:
Wow, looks great. Is that URL stable? Can we link
On Mon, 2006-01-16 at 14:43 -0500, Tom Lane wrote:
Simon Riggs [EMAIL PROTECTED] writes:
For HJ we write each outer tuple to its own file-per-batch in the order
they arrive. Reading them back in preserves the original ordering. So
yes, caution required, but I see no difficulty, just
Simon Riggs [EMAIL PROTECTED] writes:
Sure hash table is dynamic, but we read all inner rows to create the
hash table (nodeHash) before we get the outer rows (nodeHJ).
But our idea of the number of batches needed can change during that
process, resulting in some inner tuples being initially
On Monday 16 January 2006 10:51, Tom Lane wrote:
Neil Conway [EMAIL PROTECTED] writes:
I don't think it would be all that painful. There would be no need to
convert the entire source tree to use proper Doxygen-style comments in
one fell swoop: individual files and modules can be converted
PG Build Farm wrote:
The PGBuildfarm member firefly had the following event on branch HEAD:
Failed at Stage: Check
The snapshot timestamp for the build that triggered this notification
is: 2006-01-17 03:27:00
The specs of this machine are:
OS: UnixWare / 7.1.4
Arch: i386
Comp: cc
Tom Lane [EMAIL PROTECTED] writes:
Why would we continue to dynamically build the hash table after the
start of the outer scan?
The number of tuples written to a temp file might exceed what we want to
hold in memory; we won't detect this until the batch is read back in,
and in that case
Bruce Momjian pgman@candle.pha.pa.us writes:
Wow, looks great. Is that URL stable? Can we link to it from the
PostgreSQL developers page?
The thing seems to have only the very vaguest grasp on whether it is
parsing C or C++ ... or should I say that it is convinced it is parsing
C++ despite
Greg Stark [EMAIL PROTECTED] writes:
For a hash aggregate would it be possible to rescan the original table
instead of spilling to temporary files?
Sure, but the possible performance gain is finite and the possible
performance loss is not. The original table could be an extremely
expensive
On Mon, 2006-01-16 at 20:02 -0500, Tom Lane wrote:
Simon Riggs [EMAIL PROTECTED] writes:
Sure hash table is dynamic, but we read all inner rows to create the
hash table (nodeHash) before we get the outer rows (nodeHJ).
But our idea of the number of batches needed can change during that
Title: RE: [GENERAL] [PATCH] Better way to check for getaddrinfo function.
That was very much situation specific.
But the bottomline is the default test does not include netdb.h in the test code.
So, pg uses getaddrinfo.c.And the getaddrinfo.c does not work for me.
Ipv6 client
58 matches
Mail list logo