The affected ports are the ones with gettext as a run-dependency
according to ports/INDEX-7 (5007 of them) and the ones with USE_GETTEXT
in Makefile (29 of them).
PR: ports/124340
Submitted by: edwin@
Approved by: portmgr (pav)
This includes a bunch of security fixes: CVE-2007-6067, CVE-2007-4772,
CVE-2007-6601, CVE-2007-6600 and CVE-2007-4769.
Security: http://www.postgresql.org/about/news.905
A vulnerability allows suppressing the normal checks that a SQL
function returns the data type it's declared to do. These errors can
easily be exploited to cause a backend crash, and in principle might
be used to read database content that the user should not be able to
access. [CVE-2007-0555]
The release includes a set of other fixes as well. Please see the
release information at
http://www.postgresql.org/docs/7.3/static/release.html#RELEASE-7-3-18
Security: http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2007-0555
The PostgreSQL Global Development Group today released versions 8.1.4, 8.0.8,
7.4.13 and 7.3.15. This is an urgent update to close a security hole which
can permit a SQL injection attack on some applications running PostgreSQL.
Users are urged to apply the update as soon as reasonably possible. Since the
update affects client functionality, most driver projects will be updating
this week as well.
Because the security issue involved is complex, we have added a section in
Techdocs to explain it: http://www.postgresql.org/docs/techdocs.52. Please
read this first before applying the updates.
Also, fix rc_subr startup problems on FreeBSD-7.x.
Security: http://www.postgresql.org/docs/techdocs.50
PR: ports/95154
We have not checked for this KEYWORD for a long time now, so this
is a complete noop, and thus no PORTREVISION bump. Removing it at
this point is mostly for pedantic reasons, and partly to avoid
perpetuating this anachronism by copy and paste to future scripts.
A critical fix repairs an error in ReadBuffer that can cause data loss
due to overwriting recently-added pages. This applies to the 8.1 and
8.0 branches on all platforms.
Note that this update might require a reindex of textual columns under
certain conditions; please see UPDATING.
Other fixes included are:
-- Character string locale comparison bug. This may require a REINDEX
on text column indexes in some locales, such as Hungarian.
-- Prevent accidental changes of locale by plperl
-- Two fixes for Japanese encodings
-- Two fixes for COPY CSV
-- Fixes for functions returning RECORD
-- Fixes to autovacuum, dblink and pgcrypto
"start" when booting, since there's no need waste time checking for
running processes when the OS is starting up.
Bumping portrevision.
PR: 90884
Submitted by: Victor Snezhko <snezhko@indorsoft.ru>
installed from ports. The base heimdal distribution installs libraries
that have no depenency information. While this is quite correct, it
means that each library that links with libpq.so must also know if
libpq.so is linked with libkrb.so et al. Problem is, there's no good
way to get this information (pg_config has a --libs option starting at
version 8.1) and all ports using postgresql must be changed to make it
possible to link with a libpq.so that was configured to use the
Kerberos implementation installed in /usr by default. Hence, we
require one of the ports (heimdal or krb5) if postgresql is to be
linked with Kerberos. At least for now, until we can fix this in some
better way.
Also, if MIT Kerberos (security/krb5) is installed, users should
ideally remove the base heimdal installation so linkers will not pick
it up in preference to the krb5 libs (base heimdal has higher version
numbers than krb5 port).
PR: 80869, 88098, 85178
Migration to version 7.3.11
A dump/restore is not required for those running 7.3.X. However, if you
are upgrading from a version earlier than 7.3.10, see the release notes
for 7.3.10.
__________________________________________________________________
Changes
* Fix error that allowed "VACUUM" to remove ctid chains too soon, and
add more checking in code that follows ctid links
This fixes a long-standing problem that could cause crashes in very
rare circumstances.
* Fix CHAR() to properly pad spaces to the specified length when
using a multiple-byte character set (Yoshiyuki Asaba)
In prior releases, the padding of CHAR() was incorrect because it
only padded to the specified number of bytes without considering
how many characters were stored.
* Fix missing rows in queries like UPDATE a=... WHERE a... with GiST
index on column a
* Improve checking for partially-written WAL pages
* Improve robustness of signal handling when SSL is enabled
* Various memory leakage fixes
* Various portability improvements
* Fix PL/PgSQL to handle var := var correctly when the variable is of
pass-by-reference type
please see the HISTORY file included in the Release, but a summary
consists of:
* Change encoding function signature to prevent misuse
* Change "contrib/tsearch2" to avoid unsafe use of INTERNAL function
results
* Repair race condition between relation extension and VACUUM
This could theoretically have caused loss of a page's worth of
freshly-inserted data, although the scenario seems of very low
probability. There are no known cases of it having caused more than
an Assert failure.
Security: http://www.postgresql.org/about/news.315
add support to select login class for running postgresql [2].
The new startup script, using rc.subr, is now installed for all
versions of postgresql. Bump portrevisions, since startup script is
modified.
PR: 78630 [1]
Submitted by: Vivek Khera [1]
Submitted by: Brian B. [2]
Approved by: seanc (implicit)
installed, the patched gram.y file would not be used and the security
patch would be a no-op. Also, I've had reports of compilation errors
related to bison.
Since checking for the correct version of bison is hard and error
prone, I'm doing what the postgresql distribution does - patching the
yacc:ed .c file to get rid of the building dependency.
Bumping portrevision of -server.
Pointy hat to: me
Noticed by: Mike Harding and others
Security: http://www.vuxml.org/freebsd/6b4b0b3f-8127-11d9-a9e7-0001020eed82.html
Approved by: seanc (implicit)
Prevent overrunning a heap-allocated buffer if more than 1024
parameters to a refcursor declaration are specified. This is a
minimally-invasive fix for the buffer overrun.
Define LATEST_LINK to avoid package name clashes between the different
branches of PostgreSQL. [1] (Since postgresql-tcltk is hardwired to
branch 7.4, keep its LATEST_LINK to a generic value.)
Set UNIQUENAME and let it be the same for server & client, so each
branch's ports will share the same options file. This adds some no-op
knobs to the -client port, but IMO it is better this way.
Add space inside paranthesis in OSVERSION conditional to work around
(ancient) make bug. [2]
Remove the Rendez-Vouz knob for 8.0 since I can't find the software
needed to even compile it on FreeBSD.
Bump portrevision (for -server only).
Noted by: kris [1]
PR: ports/77530 [2]
Security: http://www.vuxml.org/freebsd/6b4b0b3f-8127-11d9-a9e7-0001020eed82.html
Approved by: seanc (mentor)
the "LOAD" option, the PostgreSQL Global Development Group is
announcing the release of new versions of PostgreSQL.
Update to 7.3.9, 7.4.7 & 8.0.1.
Take the opportunity to reset PORTREVISION of slave ports.
Back out name change of startup script. The new script uses rc.subr(8),
and as such also uses rcorder(8). But, rcorder does not exist in FreeBSD
4.x. Hence rename the script it back to the top of the directory
list. [1]
The periodic script should of course be executable. [2]
[1] Noted by Niels Chr. Bank-Pedersen <ncbp at bank-pedersen dot dk>
[2] Noted by Fritz Heinrichmeyer <fritz.heinrichmeyer at fernuni-hagen dot de>
following knobs can be used by ports depending on PostgreSQL:
# USE_PGSQL - Add PostgreSQL client dependency.
# If no version is given (by the maintainer via the port or
# by the user via defined variable), try to find the
# currently installed version. Fall back to default if
# necessary (PostgreSQL-7.4 = 74).
# DEFAULT_PGSQL_VER
# - PostgreSQL default version. Can be overridden within a port.
# Default: 74.
# WANT_PGSQL_VER
# - Maintainer can set an arbitrary version of PostgreSQL by
# using it.
# BROKEN_WITH_PGSQL
# - This variable can be defined if the ports doesn't support
# one or more versions of PostgreSQL.
Note that a periodic script is installed that will, by default, vacuum
all databases nightly. You may want to tweak the settings, see the
script at ${LOCALBASE}/etc/periodic/daily/502.pgsql
To run PostgreSQL at boot time, set postgresql_enable=yes in /etc/rc.conf.
Note
PR: 75344
Approved by: portmgr@ (kris), ade & sean (mentors)
Note that none of these ports are (yet) hooked into the tree,
and will not compile unless you set a specific environmental
variable. This should be warning enough to leave well alone
for now :)
Submitted by: maintainer
advisories in http://www.postgresql.org/news/234.html
Note that postgresql 7.2.x is NOT being updated here since it is
due for termination real soon now.
Submitted by: maintainer, also referenced in ports/73142 (no patch)
* Revert erroneous changes in rule permissions checking
* Repair incorrect order of operations in GetNewTransactionId()
* Ensure configure selects -fno-strict-aliasing even when an
external value for CFLAGS is supplied
* Make pg_restore handle 64-bit off_t correctly
* Make contrib/dblink not assume that local and remote type OIDs
match (Joe)
* Quote connectby()'s start_with argument properly (Joe)
* Don't crash when a rowtype argument to a plpgsql function is NULL
* Avoid generating invalid character encoding sequences in corner
cases when planning LIKE operations
* Ensure text_position() cannot scan past end of source string in
multibyte cases (Korea PostgreSQL Users' Group)
* Fix index optimization and selectivity estimates for LIKE
operations on bytea columns (Joe)
This version, as with most minor versions, does not require
a dump/reload to put into place.
PR: ports/63698
Submitted by: Palle Girgensohn <girgen@pingpong.net>
LOCALBASE to find krb5-config in the MIT case.
Prodded by: Our friendly folks at the ISC who want to use FreeBSD's
PostgreSQL ports instead of an unmanaged build from a tarball.
Just realized that after the repo-copy, the MD5_FILE pointer for
subports was never updated.
PR: ports/61193
Submitted by: Palle Girgensohn <girgen@pingpong.net>
