Skip to content

Wöchentlicher PostgreSQL Newsletter - 19. September 2010

Der Originalartikel befindet sich unter:

== Wöchentlicher PostgreSQL Newsletter - 19. September 2010 ==

Wenn es zwischendurch keine Katastrophe gibt, wird PostgreSQL 9.0.0,
welches unter anderem eine eingebaute Replikation enthält,
am Montag, dem 20. September 2010 veröffentlicht.

Die Registrierung für Frühbucher für JDCon West 2010 ist jetzt möglich.

Surge 2010, die Konferenz über Skalierbarkeit und Performance, findet
in Baltimore, MD, USA vom 30. September bis zum 1. Oktober 2010 statt.

== PostgreSQL Produkt Neuigkeiten ==

Another PostgreSQL Diff Tool 2.0 Beta 2, ein Werkzeug um Datenbankdiffs
zu erstellen, ist erschienen.

Benetl 3.6, ein ETL Werkzeug für Dateien, die PostgreSQL
nutzen, ist erschienen.

phpPgAdmin 5.0-beta1, ein webbasiertes Administrationstool für
PostgreSQL, ist erschienen.

== PostgreSQL Jobs im September ==

== PostgreSQL Lokal ==

Surge 2010, die Konferenz über Skalierbarkeit und Performance, findet
in Baltimore, MD, USA vom 30. September bis zum 1. Oktober 2010 statt.

Der Call for Papers für den PGDay.EU 2010 am 6.-8. Dezember in
Stuttgart ist nun eröffnet.

== 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

== Reviews ==

- Abhijit Menon-Sen reviewed the "Our CLUSTER implementation is

  pessimal" patch and found two chunks that did not apply cleanly,
  along with some possible bit rot on copy_heap_data().

- KaiGai Kohei reviewed Peter Eisentraut's patch to extend ALTER TYPE.

- ITAGAKI Takahiro reviewed the SQL/MED(FDW) patch and found some

  unused new types, a missing error check, a place to improve an error
  message, and some places to improve consistency in type checks
  between TABLE and FOREIGN TABLE.

- Tom Lane reviewed Sushant Sinha's url-part extention for text


- Hitoshi Harada reviewed Pavel Stehule's WIP patch for median() and

  percentile(), and found that it left out changes to contrib, had
  some cosmetic coding style issues, has a possible new out-of-memory
  condition due to the extra tuplesorts added, contained neither
  documentation nor new regression tests, doesn't work in windowing
  function context.

== Applied Patches ==

Heikki Linnakangas committed:

- In pgsql/src/backend/utils/init/postinit.c, process options from the
  startup packed in walsender.  Only few options make sense for
  walsender, but for example application_name and client_encoding do.
  We still don't apply per-role settings from pg_db_role_setting,
  because that would require connecting to a database to read the
  table.  Fujii Masao

- In pgsql/src/include/replication/walreceiver.h, remove prototype for
  non-existent function from walreceiver.h.  Tidy up by separating
  prototypes for functions in walreceiver.c and walreceiverfuncs.c
  with comments.

- In pgsql/src/backend/port/unix_latch.c, add a comment noting that
  the owner_pid test in OwnLatch is just a sanity check, per request
  by Jeff Davis.

- In pgsql/src/bin/pg_ctl/pg_ctl.c, don't warn about an in-progress
  online backup, when we're recovering from an online backup instead
  of performing one.  pg_ctl can detect that by checking if
  recovery.conf exists.  Backup label file is renamed away early in
  recovery, so the window where backup label exists during recovery is
  normally very small, but you can run into it e.g if restore_command
  is set incorrectly and the startup process never finds even the
  first WAL segment containing the checkpoint record to start recovery
  from.  Fujii Masao with comments by me.

- In pgsql/src/backend/replication/walsender.c, oops, the timeout
  argument to WaitLatchOrSocket is in microseconds, not milliseconds.

- In pgsql/src/backend/replication/walsender.c, don't call OwnLatch
  while holding a spinlock.  OwnLatch can elog() under some "can't
  happen" scenarios, and spinlocks should only be held for a few
  instructions anyway.  As pointed out by Fujii Masao.

- Simplify Windows implementation of latches.  There's no need to keep
  a dynamic pool of event handles, we can permanently assign one for
  each shared latch.  Thanks to that, we no longer need a separate
  shared memory block for latches, and we don't need to know in
  advance how many shared latches there is, so you no longer need to
  remember to update NumSharedLatches when you introduce a new latch
  to the system.

- Use a latch to make startup process wake up and replay immediately
  when new WAL arrives via streaming replication.  This reduces the
  latency, and also allows us to use a longer polling interval, which
  is good for energy efficiency.  We still need to poll to check for
  the appearance of a trigger file, but the interval is now 5 seconds
  (instead of 100ms), like when waiting for a new WAL segment to
  appear in WAL archive.

- In pgsql/src/backend/access/transam/xlog.c, fix two typos in
  comments, spotted by Fujii Masao and Thom Brown.

Robert Haas committed:

- In pgsql/doc/src/sgml/catalogs.sgml, elaborate on what gets stored
  in pg_authid.rolpasswd.  Also, add cross-reference from
  pg_shadow.passwd to pg_authid.rolpasswd and fix a bit of markup I
  muffed in my previous commit.  Per discussion with Josh Kupershmidt.

- In pgsql/src/backend/commands/comment.c, remove duplicated code left
  behind by my recent refactoring of comment.c.  These checks are also
  present in objectaddress.c, so there's no need to recheck here.

- In pgsql/doc/src/sgml/catalogs.sgml, move pg_db_role_setting docs to
  correct place in alphabetical order.

Tom Lane committed:

- Fix join-removal logic for pseudoconstant and outerjoin-delayed
  quals.  In these cases a qual can get marked with the removable rel
  in its required_relids, but this is just to schedule its evaluation
  correctly, not because it really depends on the rel.  We were
  assuming that, in effect, we could throw away all quals so marked,
  which is nonsense.  Tighten up the logic to be a little more
  paranoid about which quals belong to the outer join being considered
  for removal, and arrange for all quals that don't belong to be
  updated so they will still get evaluated correctly.  Also fix
  another problem that happened to be exposed by this test case, which
  was that make_join_rel() was failing to notice some cases where a
  constant-false qual could be used to prove a join relation empty.
  If it's a pushed-down constant false, then the relation is empty
  even if it's an outer join, because the qual applies after the outer
  join expansion.  Per report from Nathan Grange.  Back-patch into

- In pgsql/doc/src/sgml/release-9.0.sgml, add a compatibility note
  about plpgsql's treatment of SELECT INTO rec.fld when fld is of
  composite type.  Per discussion of bug #5644 from Valentine

- Fix two new-in-9.0 bugs in hstore.  There was an incorrect Assert in
  hstoreValidOldFormat(), which would cause immediate core dumps when
  attempting to work with pre-9.0 hstore data, but of course only in
  an assert-enabled build.  Also, ghstore_decompress() incorrectly
  applied DatumGetHStoreP() to a datum that wasn't actually an hstore,
  but rather a ghstore (ie, a gist signature bitstring).  That used to
  be harmless, but could now result in misbehavior if the hstore
  format conversion code happened to trigger.  In reality, since
  ghstore is not marked toastable (and doesn't need to be), this
  function is useless anyway; we can lobotomize it down to returning
  the passed-in pointer.  Both bugs found by Andrew Gierth, though
  this isn't exactly his proposed patch.

- In pgsql/doc/src/sgml/plpgsql.sgml, fix bad grammar.

- Stamp 9.0 release notes with expected release date; also some
  last-minute copy-editing.

- In pgsql/src/backend/access/transam/README, add some documentation
  about how we WAL-log filesystem actions.  Per a question from Robert

- In pgsql/src/backend/parser/analyze.c, give a suitable HINT when an
  INSERT's data source is a RowExpr containing the same number of
  columns expected by the insert.  This suggests that there were extra
  parentheses that converted the intended column list into a row
  expression.  Original patch by Marko Tiikkaja, rather heavily
  editorialized by me.

- In pgsql/src/backend/commands/copy.c, make sure we wait for
  protocol-level EOF when ending binary COPY IN.  The previous coding
  just terminated the COPY immediately after seeing the EOF marker (-1
  where a row field count is expected).  The expected CopyDone or
  CopyFail message just got thrown away later, since we weren't in
  COPY mode anymore.  This behavior complicated matters for the JDBC
  driver, and arguably was the wrong thing in any case since a
  CopyFail message after the marker wouldn't be honored.  Note that
  there is a behavioral change here: extra data after the EOF marker
  was silently ignored before, but now it will cause an error.  Hence
  not back-patching, although this is arguably a bug.  Per report and
  patch by Kris Jurka.

- Fix several broken $PostgreSQL$ keywords.  Noted while experimenting
  with Magnus's script to remove these.

- In pgsql/src/backend/utils/mb/Unicode/, replace
  last remaining $Id$ with $PostgreSQL$.

Peter Eisentraut committed:

- Translation updates for 9.0.0.

Magnus Hagander committed:

- In pgsql/src/backend/postmaster/postmaster.c, treat exit code 128
  (ERROR_WAIT_NO_CHILDREN) as non-fatal on Win32, since it can happen
  when a process fails to start when the system is under high load.
  Per several bug reports and many peoples investigation.  Back-patch
  to 8.4, which is as far back as the "deadman-switch" for shared
  memory access exists.

Marc Fournier committed:

- Tag v9.0.0 ... the big day approaches.

Bruce Momjian committed:

- In pgsql/src/backend/access/heap/README.HOT, update HOT README about
  when single-page vacuums happen.

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Teodor Sigaev sent in another revision of the patches to add
K-Nearest-Neighbor support to GiST indexes.

Heikki Linnakangas sent in two revisions of a patch using the new
latch facility to reduce walreceiver latency.

Hitoshi Harada sent in two revisions of a patch to allow top-level DML

KaiGai Kohei sent in another revision of the security label patch.

Kevin Grittner sent in another revision of the patch to do
serializable snapshot isolation.

Bernd Helmle sent in another revision of the patch to fix a mismatch
between ALTER TABLE ... ALTER COLUMN and table inheritance.

Fujii Masao sent in two more revisions of the synchronous replication

ITAGAKI Takahiro sent in a patch, inspired by Joseph Adams's Google
Summer of Code work, to add JSON.  Joseph Adams sent in a patch based
on it.

SAKAMOTO Masahiko sent in three revisions of a patch to add FDW
(foreign data wrapper) support, which is part of SQL/MED (Management
of External Data).

Peter Eisentraut sent in another WIP patch to implement per-column

Fujii Masao sent in a patch to adjust signal handling during recovery.

Fujii Masao sent in two revisions of a patch to add a user-callable
function to trigger failover in replication.

Simon Riggs sent in a WIP patch which uses latches to adjust behavior
of the startup process.

Peter Eisentraut sent in another revision of the patch to extend ALTER
TYPE for composite types.

Fujii Masao sent in a patch to add a new function to libpq which is
used to send an ACK back from walreceiver.

Kevin Grittner sent in another revision of the patch to do
serializable row locks.

Robert Haas sent in a patch to add a pg_opfamily_is_visible()


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