Re: [HACKERS] Why O_SYNC is faster than fsync on ext3

2004-03-21 Thread Manfred Spraul
Yusuf Goolamabbas wrote: I sent this to Bruce but forgot to cc pgsql-hackers, The patches are likely to go into 2.6.6. People interested in extremely safe fsync writes should also follow the IDE barrier thread and the true fsync() in Linux on IDE thread Actually the most interesting part of

Re: [HACKERS] Why O_SYNC is faster than fsync on ext3

2004-03-21 Thread Tom Lane
Yusuf Goolamabbas [EMAIL PROTECTED] writes: Bruce, haven't followed the thread completely. Accessing the web archive is slow from Hong Kong but I just wanted to point you to this lkml post which shows why O_SYNC is much faster than fsync (at least on ext3)

Re: [HACKERS] Unbalanced Btree Indices ...

2004-03-21 Thread Tom Lane
Marc G. Fournier [EMAIL PROTECTED] writes: Pointers to docs that I'm not finding most acceptable ... There aren't any, because at the time the 7.4 docs were being prepared, we didn't have enough field experience to know whether its version of index compaction would eliminate the practical need

Re: [HACKERS] [pgsql-hackers-win32] Sync vs. fsync during

2004-03-21 Thread Sailesh Krishnamurthy
Kevin == Kevin Brown [EMAIL PROTECTED] writes: The bigger problem though with this is that it makes the problem of list overflow much worse. The hard part about shared memory management is not so much that the available space is small, as that the available space is fixed

[HACKERS] execute command tag including affected rows count

2004-03-21 Thread Kris Jurka
Would it be possible to have the command completion tag for EXECUTE return the affected row count? The JDBC API says you should be able to get the affected row count for any statement and I'm working on integrating a patch that transforms regular statements to server side prepared statements,

Re: [HACKERS] Unbalanced Btree Indices ...

2004-03-21 Thread Arthur Ward
Marc G. Fournier [EMAIL PROTECTED] writes: Pointers to docs that I'm not finding most acceptable ... There aren't any, because at the time the 7.4 docs were being prepared, we didn't have enough field experience to know whether its version of index compaction would eliminate the practical

[HACKERS] Reporting errors inside plpgsql/SPI queries

2004-03-21 Thread Tom Lane
I've been fooling with adding a report of the executing query to the CONTEXT stack when an error happens within a query submitted by a plpgsql function. Since plpgsql submits all its queries through SPI, the most convenient place to do this is in spi.c, and so the behavior will also apply to

Re: [HACKERS] Syntax error reporting (was Re: [PATCHES] syntax error position CREATE FUNCTION bug fix)

2004-03-21 Thread Tom Lane
Fabien COELHO [EMAIL PROTECTED] writes: I agree with you that both reports should not look the same. The good news is that they already do not look the same, thanks to the CONTEXT information. Right, but you quite properly didn't like my quick-hack to psql that assumes that the presence of

Re: [HACKERS] Reporting errors inside plpgsql/SPI queries

2004-03-21 Thread Joe Conway
Tom Lane wrote: Although this is quite reasonable for queries submitted by user-written C functions, I'm worried that plpgsql programmers will be confused because they've never heard of SPI. I toyed with saying SQL query instead, but that seems pretty nearly content-free ... it doesn't

Re: [HACKERS] execute command tag including affected rows count

2004-03-21 Thread Neil Conway
On 21-Mar-04, at 11:39 AM, Kris Jurka wrote: Would it be possible to have the command completion tag for EXECUTE return the affected row count? Yes, this has been suggested before -- for example: http://www.mail-archive.com/[EMAIL PROTECTED]/msg26787.html I agree it would be a good idea,

Re: [HACKERS] 7.4.2 Build broken on (Sparc) Solaris 7 and 8

2004-03-21 Thread Jim Seymour
Bruce Momjian [EMAIL PROTECTED] wrote: [snip] OK, patch attached and applied. It adds configure tests for the 5-arg version of getpwuid_r() and properly falls back to the Posix draft version you have on Solaris. Seems Solaris 9 also still has the draft version. [snip] Well, yes and no.

[HACKERS] pg_autovacuum next steps

2004-03-21 Thread Matthew T. O'Connor
Lately I have been thinking about the next steps for the pg_autovacuum daemon. I have written up a document that describes what I'm planning to do next. Please read the attached and response as I would really like some feedback. Thanks, Matthew O'Connor pg_autovacuum_v2_writeup.rtf

Re: [HACKERS] execute command tag including affected rows count

2004-03-21 Thread Tom Lane
Neil Conway [EMAIL PROTECTED] writes: On 21-Mar-04, at 11:39 AM, Kris Jurka wrote: Would it be possible to have the command completion tag for EXECUTE return the affected row count? Yes, this has been suggested before -- for example: http://www.mail-archive.com/[EMAIL

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Tom Lane
Matthew T. O'Connor [EMAIL PROTECTED] writes: Lately I have been thinking about the next steps for the pg_autovacuum daemon. I have written up a document that describes what I'm planning to do next. Please read the attached and response as I would really like some feedback. [ rtf document

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Markus Bertheau
, 22.03.2004, 00:12, Tom Lane : Please repost in some less proprietary format. Plain text is generally considered the thing to use on this list. -- Markus Bertheau [EMAIL PROTECTED] pg_autovacuum Version 2 Design Document: Exec Summary: pg_autovacuum was initially released as a contrib

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Peter Eisentraut
Matthew T. O'Connor wrote: Lately I have been thinking about the next steps for the pg_autovacuum daemon. I have written up a document that describes what I'm planning to do next. Please read the attached and response as I would really like some feedback. I think these configuration issues

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Christopher Kings-Lynne
I think these configuration issues will become a lot easier if you make the autovacuum daemon a subprocess of the postmaster (like, say, the checkpoint process). Then you have access to a host of methods for storing state, handling configuration, etc. Yeah - why delay making it a backend

Re: [HACKERS] execute command tag including affected rows count

2004-03-21 Thread Kris Jurka
On Sun, 21 Mar 2004, Tom Lane wrote: On 21-Mar-04, at 11:39 AM, Kris Jurka wrote: Would it be possible to have the command completion tag for EXECUTE return the affected row count? However, does this really solve Kris' problem? JDBC generally likes to think that it works with old

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Matthew T. O'Connor
On Sun, 2004-03-21 at 20:31, Christopher Kings-Lynne wrote: I think these configuration issues will become a lot easier if you make the autovacuum daemon a subprocess of the postmaster (like, say, the checkpoint process). Then you have access to a host of methods for storing state,

Re: [HACKERS] execute command tag including affected rows count

2004-03-21 Thread Bruce Momjian
Neil Conway wrote: On 21-Mar-04, at 11:39 AM, Kris Jurka wrote: Would it be possible to have the command completion tag for EXECUTE return the affected row count? Yes, this has been suggested before -- for example: http://www.mail-archive.com/[EMAIL PROTECTED]/msg26787.html I

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Matthew T. O'Connor
On Sun, 2004-03-21 at 18:12, Tom Lane wrote: Matthew T. O'Connor [EMAIL PROTECTED] writes: [ rtf document ] Please repost in some less proprietary format. Plain text is generally considered the thing to use on this list. I don't think RTF is proprietary but I should have just posted

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Gavin Sherry
On Sun, 21 Mar 2004, Matthew T. O'Connor wrote: On Sun, 2004-03-21 at 20:31, Christopher Kings-Lynne wrote: I think these configuration issues will become a lot easier if you make the autovacuum daemon a subprocess of the postmaster (like, say, the checkpoint process). Then you have

Re: [HACKERS] pg_autovacuum next steps

2004-03-21 Thread Bruce Momjian
B) Perhaps people like the idea of it being a client app (I don't think so.) I'd like to see it as part of the backend. C) Most importantly, I'm not backend hacker. If someone wants to do the initial work of getting it running as a backend process, I can take it from there. A

Re: [HACKERS] 7.4.2 Build broken on (Sparc) Solaris 7 and 8

2004-03-21 Thread Bruce Momjian
Jim Seymour wrote: Bruce Momjian [EMAIL PROTECTED] wrote: [snip] OK, patch attached and applied. It adds configure tests for the 5-arg version of getpwuid_r() and properly falls back to the Posix draft version you have on Solaris. Seems Solaris 9 also still has the draft version.

Re: [PATCHES] [HACKERS] UnixWare/CVS Tip/initdb.c needs to use threads

2004-03-21 Thread Bruce Momjian
Larry Rosenman wrote: The a.out (not any library) should be linked with -Kpthread (not -lpthread). This will force libthread to be linked in the right order relative to libc, libC, networking libraries, etc. In other words, the entire application either is or is not linked with threads;

Re: [HACKERS] COPY formatting

2004-03-21 Thread Karel Zak
On Fri, Mar 19, 2004 at 09:54:37AM -0500, Tom Lane wrote: Karel Zak [EMAIL PROTECTED] writes: The problem with CSV is that it will correctly work with new protocol only. Because old versions of clients are newline sensitive. Why? The client-side code doesn't have any real say over