Bugs item #841817, was opened at 2003-11-13 18:26 Message generated for change (Settings changed) made by jsquyres You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=109368&aid=841817&group_id=9368
Category: Installation >Group: 4.1 Status: Open Resolution: None Priority: 5 Submitted By: Terry Fleury (tfleury) Assigned to: John (muglerj) Summary: Uninstall fails if missing script for any uninstall phase Initial Comment: Currently, when any part of the package_uninstall subroutine fails, it returns IMMEDIATELY and none of the (possible) subsequent uninstall scripts for the package get executed. This implies that a package author must write ALL of the uninstall scripts, even if those scripts do nothing. In the past in OSCAR, we have tried to use a "no configuration file implies default values" approach. I think that a package author should only have to write uninstall scripts for those parts that require it. If there isn't an uninstall script for a particular phase, then nothing happens for that phase, but it should not imply a complete failure for uninstalling a package. For example, if a package is installed only on the server, then the user should only have to write a post_server_rpm_uninstall script. However, since the current uninstall for the server comes after the uninstall for the clients, the uninstall for the server never gets executed. ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2004-11-01 16:26 Message: Logged In: YES user_id=288102 Can't update to intelligently rm RPMS until using DepMan/PackMan in image /client build. Therefore requires we "require" uninstall script. ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2004-10-27 11:03 Message: Logged In: YES user_id=288102 Confirm that this issues has been resolved. ---------------------------------------------------------------------- Comment By: Jeff Squyres (jsquyres) Date: 2003-11-25 14:57 Message: Logged In: YES user_id=11722 This should be fixed after 3.0 -- meaning that if there is no script, it should be considered success, and all the "exit 0" scripts should be removed. ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2003-11-15 19:33 Message: Logged In: YES user_id=288102 i believe the no-op scripts are sufficient for a workaround with this release so i'm moving this bug to level=5 b/c with the workaround it is not a showstopper. ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2003-11-14 12:16 Message: Logged In: YES user_id=288102 at this stage of the release it seems that the simplest fix is going to be to just put the no-op scripts for each package uninstall. it seems that in order to determine when a package is missing a script and when it doesn't have anything to do on, for instance, the clients will require some lookups (don't always have RPMS, e.g. loghost), etc. So for the sc release we'll just add uninstall scripts that simply exit with a successful status, i added a bunch of these no-op scripts last night and they should be in the 3.0b5 tarball for testing. ---------------------------------------------------------------------- Comment By: John (muglerj) Date: 2003-11-14 11:37 Message: Logged In: YES user_id=505737 I agree with your logic, but for now I think you have to go with no-op uninstall scripts. The only way we have to test for successfull uninstallation is the return value off the script. If the scipt is not there, i think we have to assume failure at this point. If we get to where we have finer granularity of the uninstall actions, we can then have a no-op success case like in the package_install(). ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2003-11-13 19:56 Message: Logged In: YES user_id=288102 As a work around (at a very least for testing) i'm going to checkin some no-op scripts that just exit(0) for success for all pkgs in tree that don't currently have an *uninstall script. ---------------------------------------------------------------------- Comment By: Thomas Naughton (naughtont) Date: 2003-11-13 18:43 Message: Logged In: YES user_id=288102 i agree that if you don't have anything to do on the clients (or server) that there should be a noop, and success case but it may be that for this release we just have a script that returns success. I will checkin a simple version of these, and do agree that this is not the "best" thing but there are several items in there that I feel are not "best" for the current capabilities w/ packageInUn but simply are a "best effort". ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=109368&aid=841817&group_id=9368 ------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://productguide.itmanagersjournal.com/ _______________________________________________ Oscar-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-devel
