summaryrefslogtreecommitdiff
path: root/autoscripts
Commit message (Collapse)AuthorAge
* r417: * dh_installinit -r: stop init script in prerm on package removal,joey2001-02-05
| | | | Closes: #84974
* r407: * suidmanager is obsolete now, and so is dh_suidmanager. Instead,joey2001-01-10
| | | | | | | | | | | | packages that contain suid binaries should include the binaries suid in the .deb, and dpkg-statoverride can override this. If this is done to a program that previously used suidmanager, though, you need to conflict with suidmanager (<< 0.50). * Made dh_suidmanager check to see if it would have done anything before. If so, it states that it is obsolete, and refer users to the man page, which now explains the situation, and then aborts the build. If it would have done nothing before, it just outputs a warning that it is an obsolete program.
* r403: * dh_installwm: Moved update-alternatives --remove call to prerm,joey2000-12-21
| | | | | Closes: #80209 * ALso guarded all update-alternatives --remove calls.
* r392: * DH_COMPAT=3 now enables the following new features which I can't justjoey2000-11-27
| | | | | | | | | | | | | | | | | | | | | | turn on by default for fear of breaking backwards compatability: - dh_makeshlibs makes the postinst/postrm call ldconfig. Closes: #77154 Patch from Masato Taruishi <taru@debian.org> (modified). If you use this, be sure dh_makeshlibs runs before dh_installdeb; many old rules files have the ordering backwards. - dh_installdeb now causes all files in /etc to be registered as conffiles. - debian/README is now supported: it is treated exactly like debian/README.Debian. Either file is installed as README.Debian in non-native packages, and now as just README in native packages. Closes: #34628 * This is really only the start of the changes for v3, so use with caution.. * dh_du has finally been removed. It has been deprecated for ages, and a grep of the archive shows that nothing is using it except biss-awt and scsh. I filed bugs on both almost exactly a year ago. Those bugs should now be raised to severity important.. * --number option (to dh_installemacsen) is removed. It has been deprecated for a while and nothing uses it. Use --priority instead.
* r382: * dh_suidregister: do not unregister on purge, since it will have ↵joey2000-11-06
| | | | | | already been unregistered then, and a warning will result.
* r370: * dh_installmodules: corrected the code added to maintainer scripts ↵joey2000-09-17
| | | | | | | | | so it does not call depmod -a. update-modules (which it always called)_ handles calling depmod if doing so is appropriate. Packages built with proir versions probably have issues on systems with non-modular kernels, and should be rebuilt. Closes: #71841
* r365: * dh_installxfonts: corrected a problem during package removal that wasjoey2000-08-17
| | | | silently neglecting to remove the fonts.dir/alias files.
* r353: * Modified all postinst script fragments to only run when called withjoey2000-07-06
| | | | | | | | "configure". I looked at the other possibilities, and I don't think any of the supported stuff should be called if the postist is called for error unwinds. Closes: #66673 * Implemented dh_clean -X, to allow specification of files to not delete, Closes: #66670
* r344: * Patch from Pedro Guerreiro to make install-docs only be called onjoey2000-04-17
| | | | configure and remove/upgrade. Closes: #62513
* r338: * Patch from Jorgen `forcer' Schaefer <forcer at mindless.com> (muchjoey2000-03-02
| | | | | | | | | | | | | | | | | modified)to make dh_installwm use new window manager registration method, update-alternatives. Closes: #52156, #34684 (latter bug is obsolete) * Fixed $dh{flavor} to be upper-case. * Deprecated dh_installemavcsen --number; use --priority instead. Also, the option parser requires the parameter be a number now. And, dh_installwm now accepts --priority, and window manager packages should start using it. * dh_installwm now behaves like a proper debhelper command, and reads debian/<package>.wm too. This is a small behavior change; filenames specified on the command line no longer apply to all packages it acts on. I can't belive this program existed for 2 years with such a glaring problem; I guess most people don't need ot register 5 wm's in 3 sub-packages. Anyway, it can handle such things now. :-) * Moved Dh_*.pm to /usr/lib/perl5/Debian/Debhelper. *big* change.
* r300: * dh_installdebconf: Modified to use new confmodule debconf library.joey1999-10-30
|
* r298: * Fixed some problems with dh_installxfonts font dirs.joey1999-10-28
|
* r295: * Corrected a path name in dh_installxfonts. Closes: #48315joey1999-10-25
|
* r294: * Fixed problem with dh_installemacsen options not working, patch fromjoey1999-10-24
| | | | | | | | | | | | | | | | | | | | | Rafael Laboissiere <rafael@icp.inpg.fr>, Closes: #47738 * Added new dh_installxfonts script by Changwoo Ryu <cwryu@dor17988.kaist.ac.kr>. Closes: #46684 I made some changes, though: - I rewrote lots of this script to be more my style of perl. - I removed all the verbisity from the postinst script fragment, since that is a clear violation of policy. - I made the postinst fail if the mkfontdir, etc commands fail, because this really makes more sense. Consider idempotency. - I moved the test to see if the font dir is really a directory into the dh_ script and out of the snippet. If the maintainer plays tricks on us, mkfontdir will blow up satisfactorally anyway. - So, the snippet is 9 lines long now, down from 20-some. - I realize this isn't following the reccommendations made in Brandon's font policy. I'll fight it out with him. :-) - In postrm fragment, used rmdir -p to remove as many parent directories as I can. - s:/usr/lib/X11/:/usr/X11R6/lib/X11/:g
* r284: * Enhanced debconf support -- the database is now cleaned up on packagejoey1999-09-29
| | | | | | purge. * Broke all debconf support off into a dh_installdebconf script. This seems conceptually a little cleaner.
* r274: * dh_installemacsen: the script fragments it generates now test for thejoey1999-09-12
| | | | | | | existance of emacs-package-install/remove before calling them. Though a strict reading of the emacsen policy indicates that such a test shouldn't be needed, there may be edge cases (cf bug 44924), where it is.
* r269: Addedjoey1999-09-07
|
* r189: Initial revisionjoey1999-08-17
|
* r128: Initial revisionjoey1999-08-17
|
* r103: Initial revisionjoey1999-08-17
|
* r11: Initial revisionjoey1999-08-17
|
* r9: Oops, it's a symlink. I see...joey1999-08-17
|
* r7: Fixed.joey1999-08-17
|
* r3: Initial revisionjoey1999-08-17