* portversion(1) also reads MOVED and trace origin change,
and, when invoked with "-v", displays the new origin.
Example:
% portversion -v screen
screen-4.0.2_2 < needs updating (port has 4.0.2_3) (=> 'sysutils/screen')
* Add "--ignore-moved" to portupgrade(1) and portversion(1).
When invoked with this option, both programs totally ignore MOVED.
If you encounter strange behaviour of these programs, try this out.
* Add IGNORE_MOVED option to pkgtools.conf.
This can be used to selectively ignore MOVED by pkgs.
See pkgtools.conf.sample for details.
* Keep the order of MOVED entries, and do not trace back to old entries.
Previously, when encounters the following entries,
editors/emacs|editors/emacs19|2004-03-20|emacs 19.x moved to a non-default port location
editors/emacs21|editors/emacs|2004-03-20|emacs 21.x moved to default port location
portupgrade traces as "editors/emacs21" -> "editors/emacs" -> "editors/emacs19".
I thought this behavior should not be what we want to, so added this change.
PR: ports/91272
Submitted by: KOMATSU Shinichiro <koma2@lovepeers.org> (maintainer)
Thank you for reporting. I also discovered this problem a few hours ago.
The source of this trouble is that the following line exists in MOVED,
lang/php4|lang/php4|2003-05-22|re-separated from www/mod_php4
where "moved from" and "moved to" is the same, but portupgrade
does not check this case and infinite loop occurs.
Fix to this problem will be in the next portupgrade release,
but in the meanwhile, whould you commit the following patch,
please?
PR: ports/91209
Submitted by: KOMATSU Shinichiro <koma2@lovepeers.org>
(1) For years, I have experienced that portupgrade sometimes
runs very slow. After some investigation, I found that
"pkgdb -aF" is invoked from portupgrade or other programs
(such as portversion) without having sufficient privilege
to modify /var/db/pkg/, and that error is not caught by the
caller program, and "pkgdb -aF" is invoked repeatedlly.
So, I fixed the following things:
- When pkgdb internally invoked from other program exits
with error, make the caller program also exit with error.
- Display the error messages of pkgdb that run internally
from other program.
When portupgrade or other programs die with the following error,
** You do not own /var/db/pkg. (use -f to force or run as root)
Command failed [exit code 1]: /usr/local/sbin/pkgdb -aFQ
please run "pkgdb -aF" as the user that owns /var/db/pkg (usually root).
(2) Make new tarball that contains the above fix and all
patches in files/. Bump PORTVERSION to 20051204 and change
MASTER_SITES.
PR: ports/89917
Submitted by: KOMATSU Shinichiro <koma2@lovepeers.org>
fix minor documentation issues with the portsclean(1) manpage.
1. In the SYNOPSIS, the -L option is excluded.
2. In the description for the -Q option, an example of the -QQ option
is not given and since the other descriptions do this,
it looks nicer and is easier to read.
PR: ports/82666
Submitted by: KOMATSU Shinichiro <koma2@lovepeers.org> (maintainer)
Pointed out by: Jason E. Hale <jhale@bluebottle.com>
every run, which also makes portversion(1) works for non-root users
again. This is done by simply having the flag file in /var/db/
(thanks to pav for the suggestion).
- While here anyway, fix some make warnings during install on FreeBSD 6.
PR: ports/81088
Reported by: many
Tested by: Yoshiaki Kasahara <kasahara@nc.kyushu-u.ac.jp>
Approved by: erwin (mentor)
portsdb(1):
- Add a -F flag, which calls make fetchindex internally to
update the ports index file.
portupgrade(1):
- Make -fP properly reinstall an installed package using a binary
package.
- Do not use a binary package when custom MAKE_ARGS or -m option is
defined, unless -PP is specified.
portsclean(1):
- -L: Make sure that a library does not shadow itself.
misc.:
- RcNG'ify the startup script.
- Do not use Object#class which is obsolete in Ruby >=1.8.
error when such configuration variable as MAKE_ARGS, typically
BEFORE_DEINSTALL is not defined.
Reported by: krion, Tsurutani Naoki <turutani@scphys.kyoto-u.ac.jp>
portupgrade:
- Make -PP/--use-packages-only work better, just as I originally
intended. Now packages are correctly identified using
pkg_info(1), "LATEST_LINK" files are also properly detected.
Related informational messages have also been improved.
It turned out that the LATEST_LINK/NO_LATEST_LINK detection was
broken long ago and hasn't been working for long.
pkg_fetch(1):
- Properly rename a "latest link" file with a full package name with
version part.
pkgtools.conf(5):
- Improve and unify the method for pattern matching in MAKE_ARGS,
BEFOREBUILD, BEFOREDEINSTALL and AFTERINSTALL variables.
This should fix some cases where it looks like MAKE_ARGS entries
are ignored.
overall:
- Update Copyright years.
- Apply massive message improvement. (I hope..)
pkgtools.conf(5):
- pkg_site_mirror(): Use pointyhat.FreeBSD.org instead of bento and
beta.
portupgrade(1):
- Indicate some pieces of information on what is going on in the
process title.
portversion(1):
- Make a generated script (with -c) accept additional arguments for
portupgrade(1).
portsdb(1):
- COMMENTFILE is deprecated. Weird. I thought I committed these
changes long long ago.
Submitted by: Jean-Yves Lefort <jylefort@brutele.be>.
pkgdb(1):
- Add a missing comma and unbreak the cyclic dependency
fixer.
Submitted by: ITO Tsuyoshi <tsuyoshi@is.s.u-tokyo.ac.jp>
portupgrade(1):
- Update man page. Running pkgdb -F is no longer mandatory unless so
directed by the pkgtools.
For the record, the previous tarball (pkgtools-20021101.tar.bz2) was
downloaded approximately 10,000 times in one month, 32% from the
United States, 17% from Japan, 6% from Germany, and 4% from
Canada. (followed by AU, NL, FR, GB, RU, SE, PL, ...)
pkgtools.conf(5):
- Add new variables: USE_PKGS and USE_PKGS_ONLY.
Requested by: Jeff Penn <jeff@jrpenn.demon.co.uk>
overall:
- Recognize INDEXFILE defaulting to 'INDEX'.
- Introduce a "fixme" cookie file not to invoke pkgdb -aF unnecessarily.
- Recognize the Hungarian and Portuguese language prefixes.
Approved by: kris (implicitly)
* No need to recurse @pkgdep search, really.
* Support the latest pkg_tarup.
* Backup +REQUIRED_BY a bit earlier, just in case.
* If it fails to copy the new +REQUIRED_BY to the package's dbdir,
emit a message and save the file in TMPDIR.
