Skip to content

Wöchentlicher PostgreSQL Newsletter - 17. Juli 2011

Der Originalartikel befindet sich unter:

== Wöchentlicher PostgreSQL Newsletter - 17. Juli 2011 ==

PostgreSQL 9.1 Beta 3 ist erschienen. Testen!

PGDay Porto Alegre findet am 19. August 19 2011
in Porto Alegre, RS, Brasilien statt.

Der Abschnitt mit den Patches ist gekürzt diese Woche, um den
Newsletter zügig erscheinen zu lassen.

== PostgreSQL Produkt Neuigkeiten ==

PostgreSQL Maestro 11.7, ein Windows Administrationswerkzeug
für PostgreSQL, ist erschienen.

pg_blockinfo 0.1, ein Perl Werkzeug um PostgreSQL
Head Datendateien zu analysieren, ist erschienen.

pgbuildfarm Client 4.6 ist erschienen.

PostgreSQL Enterprise Manager Beta erschienen.

pgpool-II 3.1.0 alpha3, ein Connection Pooler und mehr,
ist erschienen.

PostgreDAC 2.6.4, ein Delphi/C++ Builder für PostgreSQL,
ist erschienen.

psqlODBC 09.00.0310, ein ODBC Treiber für PostgreSQL, ist erschienen.

== PostgreSQL Jobs im Juli ==

== PostgreSQL Lokal ==

CHAR(11), die PostgreSQL Konferenz für Clustering, Hochverfügbarkeit
und Replikation hat jetzt die Registrierung eröffnet.
Die Konferenz findet am 11. und 12. Juli 2011 in Cambridge, UK statt.

PostgreSQL Conference China 2011 findet in Guangzhou am
15. und 16. Juli 2011 statt.

PDXPUG hostet PgDay am Sonntag, dem 24. Juli 2011, einen Tag
vor der OSCON in Portland, Oregon, USA. Details hier:

Postgres Open 2011, eine Konferenz die sich auf den Umbruch der
Datenbankindustrie durch PostgreSQL konzentriert, wird vom 14. bis 16.
September 2011 in Chicago, Illinois im Westin Michigan Avenue
Hotel stattfinden.

PG-Day Denver 2011 findet am Freitag, dem 21. Oktober 2011 auf dem
Auraria Campus in der Nähe von Downtown Denver, Colorado statt.

PostgreSQL Conference West (#PgWest) findet vom 27. bis 30. September
2011 im San Jose Convention Center in Jan Jose, Kalifornen, USA statt.

PostgreSQL Conference Europe 2011 findet vom 18. bis
21. Oktober in Amsterdam statt.

pgbr findet in Sao Paulo, Brazilien, am 3. und 4. November 2011 statt.

PGConf.DE 2011 ist die Deutschsprachige PostgreSQL Konferenz
und wird am 11. November 2011 im Rheinischen Industriemuseum
in Oberhausen, Deutschland, stattfinden. Der Call for Papers ist offen.

== PostgreSQL in den News ==

Planet PostgreSQL:

Dieser wöchentliche PostgreSQL Newsletter wurde erstellt von David Fetter.

Sende Neuigkeiten und Ankündigungen bis Sonntag, 15 Uhr Pazifischer
Zeit. Bitte sende englische Beiträge an, deutsche an, italienische an, spanische an

== Reviews ==

== Angewandte Patches ==

Simon Riggs pushed:

- Reset ALTER TABLE lock levels to AccessExclusiveLock in all cases.
  Locks on inheritance parent remain at lower level, as they were
  before.  Remove entry from 9.1 release notes.

- Alter test results to comply with new ALTER TABLE behaviour.

Heikki Linnakangas pushed:

- Remove silent_mode.  You get the same functionality with "pg_ctl -l
  postmaster.log", or nohup.  There was a small issue with
  LINUX_OOM_ADJ and silent_mode, namely that with silent_mode the
  postmaster process incorrectly used the OOM settings meant for
  backend processes.  We certainly could've fixed that directly, but
  since silent_mode was redundant anyway, we might as well just remove

- Clarify that you need ActiveState perl 5.8 or later to build on

- Fix a bug with SSI and prepared transactions: If there's a dangerous
  structure T0 ---> T1 ---> T2, and T2 commits first, we need to abort
  something. If T2 commits before both conflicts appear, then it
  should be caught by OnConflict_CheckForSerializationFailure.  If
  both conflicts appear before T2 commits, it should be caught by
  PreCommit_CheckForSerializationFailure. But that is actually run
  when T2 *prepares*. Fix that in OnConflict_CheckForSerializationFailure,
  by treating a prepared T2 as if it committed already.  This is
  mostly a problem for prepared transactions, which are in prepared
  state for some time, but also for regular transactions because they
  also go through the prepared state in the SSI code for a short
  moment when they're committed.  Kevin Grittner and Dan Ports

- SSI has a race condition, where the order of commit sequence numbers
  of transactions might not match the order the work done in those
  transactions become visible to others.  The logic in SSI, however,
  assumed that it does.  Fix that by having two sequence numbers for
  each serializable transaction, one taken before a transaction
  becomes visible to others, and one after it.  This is easier than
  trying to make the the transition totally atomic, which would
  require holding ProcArrayLock and SerializableXactHashLock at the
  same time.  By using prepareSeqNo instead of commitSeqNo in a few
  places where commit sequence numbers are compared, we can make those
  comparisons err on the safe side when we don't know for sure which
  committed first.  Per analysis by Kevin Grittner and Dan Ports, but
  this approach to fix it is different from the original patch.

- There's a small window wherein a transaction is committed but not
  yet on the finished list, and we shouldn't flag it as a potential
  conflict if so. We can also skip adding a doomed transaction to the
  list of possible conflicts because we know it won't commit.  Dan
  Ports and Kevin Grittner.

- Fix one overflow and one signedness error, caused by the patch to
  calculate OLDSERXID_MAX_PAGE based on BLCKSZ. MSVC compiler warned
  about these.

- Fix two ancient bugs in GiST code to re-find a parent after page
  split: First, when following a right-link, we incorrectly marked the
  current page as the parent of the right sibling. In reality, the
  parent of the right page is the same as the parent of the current
  page (or some page to the right of it, gistFindCorrectParent() will
  sort that out).  Secondly, when we follow a right-link, we must
  prepend, not append, the right page to our list of pages to visit.
  That's because we assume that once we hit a leaf page in the list,
  all the rest are leaf pages too, and give up.  To hit these bugs,
  you need concurrent actions and several unlucky accidents.  Another
  backend must split the root page, while you're in process of
  splitting a lower-level page. Furthermore, while you scan the
  internal nodes to re-find the parent, another backend needs to again
  split some more internal pages. Even then, the bugs don't
  necessarily manifest as user-visible errors or index corruption.
  While we're at it, make the error reporting a bit better if
  gistFindPath() fails to re-find the parent. It used to be an
  assertion, but an elog() seems more appropriate.  Backpatch to all
  supported branches.

- Change the way the offset of downlink is stored in GISTInsertStack.
  GISTInsertStack.childoffnum used to mean "offset of the downlink in
  this node, pointing to the child node in the stack". It's now
  replaced with downlinkoffnum, which means "offset of the downlink in
  the parent of this node". gistFindPath() already used childoffnum
  with this new meaning, and had an extra step at the end to pull all
  the childoffnum values down one node in the stack, to adjust the
  stack for the meaning that childoffnum had elsewhere. That's no
  longer required.  The reason to do this now is this new
  representation is more convenient for the GiST fast build patch that
  Alexander Korotkov is working on.  While we're at it, replace the
  linked list used in gistFindPath with a standard List, and make
  gistFindPath() static.  Alexander Korotkov, with some changes by me.

Andrew Dunstan pushed:

- Back-patch Fix bat file quoting of %ENV from commit 19b7fac8.

- Reimplement pgbison and pgflex as perl scripts instead of bat files.
  In the process, remove almost all knowledge of individual .y and .l
  files, and instead get invocation settings from the relevant make
  files.  The exception is plpgsql's gram.y, which has a target with a
  different name. It is hoped that this will make the scripts more
  future-proof, so that they won't require adjustment every time we
  add a new .l or .y file.  The logic is also notably less tortured
  than that forced on us by the idiosyncrasies of the Windows command
  processor.  The .bat files are kept as thin wrappers for the perl

- Fix comments over eagerly c&p'd.

- Improve wording of documentation on default privileges.  Per recent
  -hackers discussion.

Tom Lane pushed:

- Remove missed reference to SilentMode.

- Fix psql's counting of script file line numbers during COPY.
  handleCopyIn incremented pset.lineno for each line of COPY data read
  from a file.  This is correct when reading from the current script
  file (i.e., we are doing COPY FROM STDIN followed by in-line data),
  but it's wrong if the data is coming from some other file.  Per bug
  #6083 from Steve Haslam.  Back-patch to all supported versions.

- Improve documentation about foreign data wrapper validator
  functions.  Modified version of a patch by Shigeru Hanada.

- Restructure foreign data wrapper chapter so it has more than one
  section.  As noted by Laurenz Albe, our SGML tools deal rather oddly
  with chapters having just one <sect1>.  Perhaps the tooling could be
  fixed, but really the design of this chapter's introduction is
  pretty bogus anyhow.  Split it into a true introduction and a
  <sect1> about the FDW functions, so that it reads better and dodges
  the lack-of-a-chapter-TOC problem.

- Make the file_fdw validator check that a filename option has been
  provided.  This was already a runtime failure condition, but it's
  better to check at validation time if possible.  Lightly modified
  version of a patch by Shigeru Hanada.

- Call FDW validator functions even when the options list is empty.
  This is useful since a validator might want to require certain
  options to be provided.  The passed array is an empty text array in
  this case.  Per suggestion by Laurenz Albe, though this is not quite
  his patch.

- Remove assumptions that not-equals operators cannot be in any
  opclass.  get_op_btree_interpretation assumed this in order to save
  some duplication of code, but it's not true in general anymore
  because we added <> support to btree_gist.  (We still assume it for
  btree opclasses, though.) Also, essentially the same logic was baked
  into predtest.c.  Get rid of that duplication by generalizing
  get_op_btree_interpretation so that it can be used by predtest.c.
  Per bug report from Denis de Bernardy and investigation by Jeff
  Davis, though I didn't use Jeff's patch exactly as-is.  Back-patch
  to 9.1; we do not support this usage before that.

- Finish disabling reduced-lock-levels-for-DDL feature.  Previous
  patch only covered the ALTER TABLE changes, not changes in other
  commands; and it neglected to revert the documentation changes.

- Reclassify replication-related GUC variables as "master" and
  "standby".  Per discussion, this structure seems more understandable
  than what was there before.  Make config.sgml and
  postgresql.conf.sample agree.  In passing do a bit of editorial work
  on the variable descriptions.

- Update 9.1 release notes to reflect commits through today.

- Update examples for string-related functions.  In the example for
  decode(), show the bytea result in hex format, since that's now the
  default.  Use an E'' string in the example for quote_literal(), so
  that it works regardless of the standard_conforming_strings setting.
  On the functions-for-binary-strings page, leave the examples as-is
  for readability, but add a note pointing out that they are shown in
  escape format.  Per comments from Thom Brown.  Also, improve the
  description for encode() and decode() a tad.  Backpatch to 9.0,
  where bytea_output was introduced.

- Tag 9.1beta3.

- Fix another oversight in logging of changes in postgresql.conf
  settings.  We were using GetConfigOption to collect the old value of
  each setting, overlooking the possibility that it didn't exist yet.
  This does happen in the case of adding a new entry within a custom
  variable class, as exhibited in bug #6097 from Maxim Boguk.  To fix,
  add a missing_ok parameter to GetConfigOption, but only in 9.1 and
  HEAD --- it seems possible that some third-party code is using that
  function, so changing its API in a minor release would cause
  problems.  In 9.0, create a near-duplicate function instead.

- Avoid listing ungrouped Vars in the targetlist of
  Agg-underneath-Window.  Regular aggregate functions in combination
  with, or within the arguments of, window functions are OK per spec;
  they have the semantics that the aggregate output rows are computed
  and then we run the window functions over that row set.  (Thus, this
  combination is not really useful unless there's a GROUP BY so that
  more than one aggregate output row is possible.) The case without
  GROUP BY could fail, as recently reported by Jeff Davis, because
  sloppy construction of the Agg node's targetlist resulted in extra
  references to possibly-ungrouped Vars appearing outside the
  aggregate function calls themselves.  See the added regression test
  case for an example.  Fixing this requires modifying the API of
  flatten_tlist and its underlying function pull_var_clause.  I chose
  to make pull_var_clause's API for aggregates identical to what it
  was already doing for placeholders, since the useful behaviors turn
  out to be the same (error, report node as-is, or recurse into it).
  I also tightened the error checking in this area a bit: if it was
  ever valid to see an uplevel Var, Aggref, or PlaceHolderVar here,
  that was a long time ago, so complain instead of ignoring them.
  Backpatch into 9.1.  The failure exists in 8.4 and 9.0 as well, but
  seeing that it only occurs in a basically-useless corner case, it
  doesn't seem worth the risks of changing a function API in a minor
  release.  There might be third-party code using pull_var_clause.

- Update some comments to clarify who does what in targetlist
  creation.  No code changes; just avoid blaming query_planner for
  things it doesn't really do.

- In planner, don't assume that empty parent tables aren't really
  empty.  There's a heuristic in estimate_rel_size() to clamp the
  minimum size estimate for a table to 10 pages, unless we can see
  that vacuum or analyze has been run (and set relpages to something
  nonzero, so this will always happen for a table that's actually
  empty).  However, it would be better not to do this for inheritance
  parent tables, which very commonly are really empty and can be
  expected to stay that way.  Per discussion of a recent
  pgsql-performance report from Anish Kejariwal.  Also prevent it from
  happening for indexes (although this is more in the nature of
  documentation, since CREATE INDEX normally initializes relpages to
  something nonzero anyway).  Back-patch to 9.0, because the ability
  to collect statistics across a whole inheritance tree has improved
  the planner's estimates to the point where this relatively small
  error makes a significant difference.  In the referenced report,
  merge or hash joins were incorrectly estimated as cheaper than a
  nestloop with inner indexscan on the inherited table.  That was less
  likely before 9.0 because the lack of inherited stats would have
  resulted in a default (and rather pessimistic) estimate of the cost
  of a merge or hash join.

- Replace errdetail("%s", ...) with errdetail_internal("%s", ...).
  There may be some other places where we should use
  errdetail_internal, but they'll have to be evaluated case-by-case.
  This commit just hits a bunch of places where invoking gettext is
  obviously a waste of cycles.

- Add an errdetail_internal() ereport auxiliary routine.  This
  function supports untranslated detail messages, in the same way that
  errmsg_internal supports untranslated primary messages.  We've
  needed this for some time IMO, but discussion of some cases in the
  SSI code provided the impetus to actually add it.  Kevin Grittner,
  with minor adjustments by me

- Use errdetail_internal() for SSI transaction cancellation details.
  Per discussion, these seem too technical to be worth translating.
  Kevin Grittner

- Improve make_subplanTargetList to avoid including Vars unnecessarily.
  If a Var was used only in a GROUP BY expression, the previous
  implementation would include the Var by itself (as well as the
  expression) in the generated targetlist.  This wouldn't affect the
  efficiency of the scan/join part of the plan at all, but it could
  result in passing unnecessarily-wide rows through sorting and
  grouping steps.  It turns out to take only a little more code, and
  not noticeably more time, to generate a tlist without such
  redundancy, so let's do that.  Per a recent gripe from HarmeekSingh

- Add temp_file_limit GUC parameter to constrain temporary file space
  usage.  The limit is enforced against the total amount of temp file
  space used by each session.  Mark Kirkwood, reviewed by Cédric
  Villemain and Tatsuo Ishii

- Further thoughts about temp_file_limit patch.  Move FileClose's
  decrement of temporary_files_size up, so that it will be executed
  even if elog() throws an error.  This is reasonable since if the
  unlink() fails, the fact the file is still there is not our fault,
  and we are going to forget about it anyhow.  So we won't count it
  against temp_file_limit anymore.  Update fileSize and
  temporary_files_size correctly in FileTruncate.  We probably don't
  have any places that truncate temp files, but fd.c surely should not
  assume that.

Peter Eisentraut pushed:

- Remove unused variable to silence compiler warning

- Set user_defined_types.data_type to null.  On re-reading the
  standard, this field is only used for distinct or reference types.

- Small documentation tweaks

- Message style tweaks

- Postgres -> PostgreSQL in error message

- Message style improvements of errmsg_internal() calls

- Fix use of unportable %m format.

- Message style improvements

- Add item about Clang support to 9.1 release notes

- Fix and clarify information schema interval_precision fields.  The
  fields were previously wrongly typed as character_data; change to
  cardinal_number.  Update the documentation and the implementation to
  show more clearly that this applies to a feature not available in
  PostgreSQL, rather than just not yet being implemented in the
  information schema.

- Remove excessively backpatched gitignore files.  These caused
  directories from future releases to appear in the backbranch tree.

- Make pg_upgrade output more consistent with project style.  Add
  errno-based output to error messages where appropriate, reformat
  blocks to about 72 characters per line, use spaces instead of tabs
  for indentation, and other style adjustments.

- Mark collation-related information schema columns as not implemented
  rather than not applicable.  They have already been implemented in

- Implement information schema interval_type columns.  Also correct
  reporting of interval precision when field restrictions are
  specified in the typmod.

- Remove README.mb.big5 and They were wildly outdated.

- Small update on suggested startup file locations.  Debian/Ubuntu
  don't have a /etc/rc.d/ directory, so add some alternative names as

- Set information_schema.routines.is_udt_dependent to NO.  It
  previously said YES, but that is incorrect.

- Set information_schema.tables.commit_action to null.  The commit
  action of temporary tables is currently not cataloged, so we can't
  easily show it.  The previous value was outdated from before we had
  different commit actions.

Alvaro Herrera pushed:

- Don't try to use a constraint name as domain name.  The bug that
  caused this to be discovered is that the code was trying to
  dereference a NULL or ill-defined pointer, as reported by Michael
  Mueller; but what it was doing was wrong anyway, per Heikki
  Linnakangas.  This patch is Heikki's suggested fix.

- Move Trigger and TriggerDesc structs out of rel.h into a new
  reltrigger.h.  This lets us stop including rel.h into execnodes.h,
  which is a widely used header.

- Add support for blocked commands in isolationtester.  This enables
  us to test that blocking commands (such as foreign keys checks that
  conflict with some other lock) act as intended.  The set of tests
  that this adds is pretty minimal, but can easily be extended by
  adding new specs.  The intention is that this will serve as a basis
  for ensuring that further tweaks of locking implementation preserve
  (or improve) existing behavior.  Author: Noah Misch

- Blind attempt at fixing isolation_tester on Win32

- Fix broken markup

- Set different deadlock_timeout on each session in new isolation
  tests.  This provides deterministic deadlock-detection ordering for
  new isolation tests, fixing the sporadic failures in them.  Author:
  Noah Misch

Magnus Hagander pushed:

- Fix typo in sslmode documentation.  Per bug #6089, noted by Sidney

- perltidy run over msvc build system

- Fix SSPI login when multiple roundtrips are required.  This fixes
  SSPI login failures showing "The function requested is not
  supported", often showing up when connecting to localhost. The
  reason was not properly updating the SSPI handle when multiple
  roundtrips were required to complete the authentication sequence.
  Report and analysis by Ahmed Shinwari, patch by Magnus Hagander

Robert Haas pushed:

- Attempt to standardize formatting of psql queries.  Most queries end
  with a backslash, but not a newline, so try to standardize on that,
  for the convenience of people using psql -E to extract queries.
  Josh Kupershmidt, reviewed by Merlin Moncure.

- Add \ir command to psql.  \ir is short for "include relative"; when
  used from a script, the supplied pathname will be interpreted
  relative to the input file, rather than to the current working
  directory.  Gurjeet Singh, reviewed by Josh Kupershmidt, with
  substantial further cleanup by me.

- Adjust OLDSERXID_MAX_PAGE based on BLCKSZ.  The value when BLCKSZ =
  8192 is unchanged, but with larger-than-normal block sizes we might
  need to crank things back a bit, as we'll have more entries per page
  than normal in that case.  Kevin Grittner

- Try to acquire relation locks in RangeVarGetRelid.  In the previous
  coding, we would look up a relation in RangeVarGetRelid, lock the
  resulting OID, and then AcceptInvalidationMessages().  While this
  was sufficient to ensure that we noticed any changes to the relation
  definition before building the relcache entry, it didn't handle the
  possibility that the name we looked up no longer referenced the same
  OID.  This was particularly problematic in the case where a table
  had been dropped and recreated: we'd latch on to the entry for the
  old relation and fail later on.  Now, we acquire the relation lock
  inside RangeVarGetRelid, and retry the name lookup if we notice that
  invalidation messages have been processed meanwhile.  Many
  operations that would previously have failed with an error in the
  presence of concurrent DDL will now succeed.  There is a good deal
  of work remaining to be done here: many callers of RangeVarGetRelid
  still pass NoLock for one reason or another.  In addition, nothing
  in this patch guards against the possibility that the meaning of an
  unqualified name might change due to the creation of a relation in a
  schema earlier in the user's search path than the one where it was
  previously found.  Furthermore, there's nothing at all here to guard
  against similar race conditions for non-relations.  For all that,
  it's a start.  Noah Misch and Robert Haas

Bruce Momjian pushed:

- Add C comment that txid_current() assigns an XID if one is not
  already assigned.

- Use clearer woring for pg_locks columns, identifying which items are
  related to lock objects.

- In docs, start window function sentence with "The asterisk (*)"
  rather than "*";  it is confusing to start a sentence with a symbol.

- Doc clarification for pg_locks columns.  Florian Pflug

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Heikki Linnakangas sent in another revision of the patch to create a
latch implementation that wakes on postmaster death on both Win32 and

Shigeru HANADA sent in a patch to add a "force not NULL" option to the
file FDW.

Shigeru HANADA and Pavel Stehule traded patches to enhance PL/pgsql's
GET DIAGNOSTICS by adding an optional STACKED keyword.

Fujii Masao sent in another revision of the patch to enable cascading

KaiGai Kohei sent in three more revisions of the patch to fix some
leaks in views.

KaiGai Kohei sent in another revision of the patch to add SECURITY
LABEL to shared database objects.


No Trackbacks


Display comments as Linear | Threaded

No comments

Add Comment

Enclosing asterisks marks text as bold (*word*), underscore are made via _word_.
E-Mail addresses will not be displayed and will only be used for E-Mail notifications.
To leave a comment you must approve it via e-mail, which will be sent to your address after submission.
Form options