ads' corner

== Angewandte Patches ==

Michael Meskes committed:

- In pgsql/src/interfaces/ecpg/preproc/README.parser, added some
  explanation about how the parser is generated, taken from an email
  by Zoltan Boszormenyi.

- In ECPG, applied patch by Zoltan Boszormenyi to fix memory leak in
  decimal handling.

Andrew Dunstan committed:

- In pgsql/src/pl/plperl/plperl.c, convert a perl array to a postgres
  array when returned by Set Returning Functions as well as +non SRFs.
  Backpatch to 8.1 where these facilities were introduced.  with a
  little help from Abhijit Menon-Sen.

Tom Lane committed:

- Fix equivclass.c's not-quite-right strategy for handling X=X
  clauses.  The original coding correctly noted that these aren't just
  redundancies (they're effectively X IS NOT NULL, assuming = is
  strict).  However, they got treated that way if X happened to be in
  a single-member EquivalenceClass already, which could happen if
  there was an ORDER BY X clause, for instance.  The simplest and most
  reliable solution seems to be to not try to process such clauses
  through the EquivalenceClass machinery; just throw them back for
  traditional processing.  The amount of work that'd be needed to be
  smarter than that seems out of proportion to the benefit.  Per bug
  #5084 from Bernt Marius Johnsen, and analysis by Andrew Gierth.

- Allow MOVE FORWARD n, MOVE BACKWARD n, MOVE FORWARD ALL, MOVE
  BACKWARD ALL in plpgsql.  Clean up a couple of corner cases in the
  MOVE/FETCH syntax.  Pavel Stehule.

- Assorted improvements in contrib/hstore.  Remove the 64K limit on
  the lengths of keys and values within an hstore.  (This changes the
  on-disk format, but the old format can still be read.)  Add support
  for btree/hash opclasses for hstore --- this is not so much for
  actual indexing purposes as to allow use of GROUP BY, DISTINCT, etc.
  Add various other new functions and operators.  Andrew (RhodiumToad)
  Gierth.

- In pgsql/contrib/hstore/hstore_op.c, fix bogus Assert, per buildfarm
  results.

- Support "samehost" and "samenet" specifications in pg_hba.conf, by
  enumerating the machine's IP interfaces to look for a match.  Stef
  Walter.

- Fix erroneous handling of shared dependencies (ie dependencies on
  roles) in CREATE OR REPLACE FUNCTION.  The original code would
  update pg_shdepend as if a new function was being created, even if
  it wasn't, with two bad consequences: pg_shdepend might record the
  wrong owner for the function, and any dependencies for roles
  mentioned in the function's ACL would be lost.  The fix is very
  easy: just don't touch pg_shdepend at all when doing a function
  replacement.  Also update the CREATE FUNCTION reference page, which
  never explained exactly what changes and doesn't change in a
  function replacement.  In passing, fix the CREATE VIEW reference
  page similarly; there's no code bug there, but the docs didn't say
  what happens.

- Make sure that GIN fast-insert and regular code paths enforce the
  same tuple size limit.  Improve the error message for
  index-tuple-too-large so that it includes the actual size, the
  limit, and the index name.  Sync with the btree occurrences of the
  same error.  Back-patch to 8.4 because it appears that the
  out-of-sync problem is occurring in the field.  Teodor Sigaev and
  Tom Lane.

- In pgsql/src/backend/postmaster/pgstat.c, fix an oversight in an
  8.3-era patch: pgstat_initstats should allow stats to be collected
  for sequences.  Report and fix by Akira Kurosawa.

- Fix a couple of issues in recent patch to print updates to
  postgresql.conf settings: avoid calling superuser() in contexts
  where it's not defined, don't leak the transient copies of
  GetConfigOption output, and avoid the whole exercise in postmaster
  child processes.  I found that actually no current caller of
  GetConfigOption has any use for its internal check of
  GUC_SUPERUSER_ONLY.  But rather than just remove that entirely, it
  seemed better to add a parameter indicating whether to enforce the
  check.  Per report from Simon and subsequent testing.

- In pgsql/src/backend/libpq/hba.c, fix assorted memory leaks in
  pg_hba.conf parsing.  Over a sufficiently large number of SIGHUP
  cycles, these would have run the postmaster out of memory.  Noted
  while testing memory-leak scenario in postgresql.conf
  configuration-change-printing patch.

Peter Eisentraut committed:

- In pgsql/doc/src/sgml/stylesheet.dsl, add tooltips to the header
  links, and make title a link to the home page based on an idea by
  Richard Huxton.

Alvaro Herrera committed:

- Ensure that a cursor has an immutable snapshot throughout its
  lifespan.  The old coding was using a regular snapshot, referenced
  elsewhere, that was subject to having its command counter updated.
  Fix by creating a private copy of the snapshot exclusively for the
  cursor.  Backpatch to 8.4, which is when the bug was introduced
  during the snapshot management rewrite.

Bruce Momjian committed:

- In pgsql/doc/src/sgml/config.sgml, document the purpose of the GUC
  listen_addresses.

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Heikki Linnakangas sent in more fixes atop earlier Hot Standby
patches.

Abhijit Menon-Sen sent in another patch to fix return_next in PL/Perl
for arrays.

ITAGAKI Takahiro sent in two more revisions of the patch expanding the
functionality of CREATE TABLE LIKE...

KaiGai Kohei sent in another revision of the large object ACL patch.

Marko (johto) Tiikkaja sent in another revision of the patch to
refactor the executor to have nodes for INSERT, UPDATE and DELETE.
This is among other things infrastructure for the Writeable CTE
feature.

Alvaro Herrera sent in missing files from the patch to add per-ROLE,
per-database GUC settings, then a patch for psql to make it display
same.

Devrim GUNDUZ sent in a doc patch which updates links to some
external projects.

Euler Taveira de Oliveira sent in a reviewed version of ITAGAKI
Takahiro's patch to include buffer usage in EXPLAIN.

Brad T. Sliger sent in a revised version of the patch to allow Unicode
codes for pretty-printing in psql.

KaiGai Kohei sent in a patch atop the ACL rework patch to eliminate
permission checks in FindConversion() and EnableDisableRule().

Alvaro Herrera sent in three more revisions of the per-ROLE,
per-database GUC patch.

Robert Leigh sent in two more revisions of the patch to add UTF8
pretty-printing to psql.

Petr (PJMODOS) Jelinek sent in another revision of the DefaulsACLs
patch and a follow-up patch with adjusted docs.

Laurenz Albe sent in a patch to add a hook for checking password
strength, per discussion.

Heikki Linnakangas sent in another patch for Hot Standby.

David Gould sent in a patch to limit the amount of memory allocated
per session.

Brendan Jurd sent in a patch for the named/mixed notation for
PL/pgsql.

ITAGAKI Takahiro sent in another revision of the buffer usage patch
for EXPLAIN.

Fujii Masao sent in another revision of the Streaming Replication
patch.

Jeff Davis sent in another revision of the mixed/named function
notation patch for PL/pgsql.

Roger Leigh sent in another revision of the Unicode pretty-printing
patch for psql, this time with a switch to change to ASCII format in
UTF-8 locales.

KaiGai Kohei sent in another revision of the ACL rework patch.

Bernd Helmle sent in a patch atop the per-ROLE, per-database GUC
settings patch which adds ALTER ROLE ... IN DATABASE functionality.

Pavel Stehule sent in another revision of the named/mixed PL/pgsql
parameter notation based in part on Brendan Jurd's and Jeff Davis's
patches.

Zoltan Boszormenyi sent in another ECPG patch for Informix
compatibility.

Zoltan Boszormenyi sent another flock of patches for dynamic cursor
names in ECPG.

Alvaro Herrera sent in a patch to remove the useless pushing of an
active snapshot on PortalStart.

Ian Small sent in a patch which does alphagram fuzzy string matching.

Joe Conway sent in a patch to fix a memory leak in dblink.