port (files/pgsql.sh.tmpl) lacks a parameter in the
restart option, causing logs to be send to stdout,
instead of the log file.
Submitted by: Fernando Schapachnik <fernando@mecon.gov.ar>
Approved by: maintainer
PR: 53142
Approved by: fjoe (implicit)
1. Optionally link with libc_r to get plpython working. [1]
2. Fix kerberos build. [2]
3. There was a duplication of some declarations. [3]
PR: ports/52851
PR: ports/51080 [2]
Submitted by: Mike Meyer <mwm@mired.org> [1]
Submitted by: Gerweck <andy@tacnode.com> [2]
Pointed out by: Mike Harding <mvh@ix.netcom.com> [3]
Submitted by: Palle Girgensohn <girgen@pingpong.net> (maintainer)
atoi('') (ex: RT and Horde). While I'm here, de-"pkg-comment"-ify. Port
revision bump.
Submitted by: Larry Rosenman <ler@lerctr.org>
Approved by: maintainer
queues and shutting down the database. Not bumping port revision, but if
you are having problems related to the above, update as necessary.
Submitted by: Larry Rosenman <ler@lerctr.org>
Fixes numerous bugs especially with various interface libraries and
pg_dump. All users are advised to upgrade. This update fixes all known
problems with the postgresql7 port. See release notes for details:
http://developer.postgresql.org/docs/postgres/release-7-3-2.html
A dump/restore is *not* required when upgrading to this version.
PR: ports/47983 [1], ports/47284 [2], ports/47808 [3]
Submitted by: maintainer [1]
Jason C. Wells [2]
Michel Oosterhof <m.oosterhof@xs4all.nl> [3]
The following buffer overruns have been identified and
addressed [in 7.2.2]:
* in handling long datetime input
* in repeat()
* in lpad() and rpad() with multibyte
* in SET TIME ZONE and TZ env var
* The compiler in -CURRENT now complains if you precede -I/usr/include
with -I... . The same applies to -I/usr/lib .
* The distribution's configure file treats the compiler's unexpected
warning message as a failure and errors out.
Maintainer notified in private mail.
- switch devel/gettext (0.11.1) on, installing full package
- flip devel/gettext-old (0.10.35) to installing only static binaries
with a "-old" suffix -- gettext-old will have its deorbit burn
sequence initiated just after 4.6-RELEASE
- fix up ports for the new world order
Reviewed by: portmgr
A note about how to install languages into a PostgreSQL database is added.
PR: ports/29916
Submitted by: Palle Girgensohn <girgen@partitur.se> (MAINTAINER)
Michal Pasternak <doc@lublin.t1.pl> (the note)
ports by defining LIB_DEPENDS for the BATCH defaults in the Makefile.
The dynamic determination of LIB_DEPENDS in configure.postgresql
remains in place for the !BATCH case and is unaffected by this.
Remove trailing whitespace and an extraneous LDCONFIG_DIRS.
Submitted by: maintainer
* Take a stab at removing the pgsql user's home directory on deinstall.
* Only remove man/manl and others when PG_PREFIX is set.
Submitted by: maintainer
checked in the fetch target. Before the patch below, make install will
fail to depend on Tcl/Tk, for instance.
Synchronize the batch and default interactive selections for optional
parts of the port.
PR: ports/27949
Submitted by: Palle Girgensohn <girgen@partitur.se> (MAINTAINER)
to make dependencies work.
Call the script pre-fetch and include Makefile.inc AFTER it has been
created.
Reported by: Disturbed FreeBSD newbie who couldn't figure out
why this didn't work as expected.
* Pass the -s option to pg_ctl, to avoid clobbering the display.
It will show error messages only.
* Echo the port name after shutdown.
No functional changes, no PORTREVISION.
* Move the call to configure.postgresql7 from pre-fetch to pre-extract, so it
won't hang while performing batch fetch operations (like portupgrade -F)
* Add some TCL related files to pkg-plist.tcl, and add a PLIST_SUB in the
Makefile to register the correct tcl version in the plist.
* Do not start postgresql if the database directory does not exist: the
startup sequence could hang because of this.
* Use the "-s" option when starting postgresql with pg_ctl, so it won't
display informational messages. Display only the port name, as do other
packages startup scripts.
Approved by: Palle Girgensohn <girgen@partitur.se>
This would cure headache like apache module linking against libpq
fail to work after rebooting.
Originally, maintainer and I want this commit to go with the
PostgreSQL updating. But we both think that the latest PostgreSQL
would not be released before FreeBSD 4.3-RELEASE ports freeze.
As a conclusion, PORTREVISION bumped.
Discussed with/Submitted by: MAINTAINER
- pkg-install used wrong uid#
- chown lib dir *after* it is created
- pkg-install created data dir where Makefile didn't
- add pkg-deinstall and remove user & group
- don't chown data dir, since we don't create it anymore
- spell and documentation fixes in the rc.d script
PR: ports/25817
Submitted by: MAINTAINER
Do not call initdb from Makefile's post-install. Users are required to
manually run initdb, maybe after configuring their preferred database
location in pgsql's user .profile. pkg_add used to simply replace some
existing data files.
Add pkg-message to tell the user about initdb.
Remove 2 empty directories from pkg-plist (pgsql/include/port/freebsd,
pgsql/include/port), since they are not included in "make package" anyway
and cause pkg_deinstall to complain about the plist.
Remove all pgsql/data contents from pkg-plist, so "make deinstall" and
pkg_delete won't clobber existing databases. Minor version upgrades can now be
performed by means of pkg_delete and "make install", without using pg_dump.
Chown %D/pgsql/lib to root:wheel to satisfy ldconfig's recent security
requirements.
Change @dirrm of pgsql to "@unexec rmdir": pkg_deinstall won't complain if
data still exists in that directory at deinstall time.
Approved by: maintainer
The DB is 64-bit capable so really only a few
bits of configuration for FreeBSD were needed.
Submitted by: Thomas Valentino Crimi <tcrimi+@andrew.cmu.edu>
default encoding.
Still the "default" default encoding is singlebyte, so this shouldn't hurt
anything but a little code increase.
Approved by: andreas (the maintainer)
and..
<quote>
This is a really small fix:
- When compiling postgresql-jdbc, the compilation process presents a
bad path to the installed jar-file.
</quote>
PR: 13838
PR: 13865
Submitted by: Palle Girgensohn <girgen@partitur.se>
Start up a /usr/ports/YEAR2000 file to provide a central location for all
software. Will be buildling a nightly run shell script to pull all the Y2K
variables from ports, so that the central list remains reasonably up to date...
Many bugfixes and cosmetic changes
Changes by Scrappy and me
My additional changes:
- had to link libpgtcl.so with the crypt library to get rid of the
pgaccess error message, that crypt is missing
- had to add -i option in the startup script, so that pgaccess is
able to connect to the postmaster process
- removed all unnecessary patches
- updated PLIST
Thanks to the postgresql developement team, who did a great job to
simplify the postgresql port, by applying the patches and making
the autoconf mechanism more consistent.
Submitted by: The Hermit Hacker <scrappy@hub.org>
background because of the -S option.
- remove the -D datadir option, it's meaningless, because the pgsql
user environment overwrites it with the PGDATA env variable.
Since this is important and might cause some headache, I mentioned
this in ~pgsql/.profile and the startup script.
Submitted by: John Fiber
- in order to access the template1 database as pgsql user, the
environment needs the DISPLAY variable set to at least ":0"
- pgaccess loads dynamically the libpgtcl.1, some symbols from
another dynamic lib are needed as well -> libpq.so.1.0
Makefile (patch-af) modified (copied from the Linux clause),
so that shared lib libpgtcl is created with proper loader flags.
for building the tcl/tk based database frontend pgaccess.
For an updated version of this tool a separate port will
follow. Please test these changes well, so that we get a
nice PostgreSQL 6.2 version up and running "b4" port freeze ;-)
Please note: when performing a migration to 6.2 and you have an existing db,
then you have to use the *new* pg_dumpall script that comes with this new
postgresql release. The INSTALL file points this out explicitely !!!
Changes:
- startup script resides in FILESDIR
- renamed it to be in sync with INSTALL file from sources
- always install this startup script over an existing, because
of the nature of the rc.d directory I can't install it
to pgsql.sh-dist, if a pgsql.sh is already presend ...
- portlint detected trailing whitespace, usage of perl with absolute
path, usage of echo instead of ECHO and plenty things of this kind
- post installation notes updated, mentioned the mailing list
- copies the html pages as well to the share/doc directory (new manual dir)
- had to update PLIST
- shortened DESCR file, to match the 24 lines
- added post build target, that reminds the admin how to proceed when
already having a database -> INSTALL file describes migration
- updated manpages
- Added missing ldconfig -m in startup file
- Modified pgsql user environment
- Make port NOPORTDOCS aware
- Add additional port docs to post-install target and updated PLIST
- recursively change owner of pgsql directory after creating .profile,
so that .profile can be modified by the pgsql user.
Hope everything is fine now.
Clean up newly imported PostgreSQL port:
- variable reorganization and cleanup
- remove NOMANCOMPRESS conditional in favour of MAN?=
- prevent import of installing user's $PATH into ${PREFIX}/pgsql/.profile
- change package name to remove 'v'
- remove some of the latent bugs left over from the postgres95 port.
to boot.
Have updated the Master Sites listing to reflect where the current port
actually resides, as well as added a few of the mirrors. Will work on
bringing the port up to date as I figure out how ports work :(
This should at least allow this port to be compiled using v1.08 now
