Emerge (supposively) keeps a database of what the latest, non-experimental versions of software are. However, it seems to latch onto the latest version marked as non-experimental, and reports that as the latest version even if the later version is marked experimental again. Steps to replicate (as of 13 November 2002): 1. "emerge search gtetrinet". It should report app-games/gtetrinet-0.4.1-r3 as the latest version. 2. Edit /usr/portage/apps-games/gtetrinet/gtetrinet-0.4.4.ebuild . Edit the KEYWORDS line so it reads "x86" and not "~x86". 3. "emerge search gtetrinet" again. Version 0.4.4 should be reported as the latest. 4. Edit the file in step #2 so it reads "~x86" again, or just "emerge sync" portage back to the way it was. 5. "emerge search gtetrinet" yet again. Version 0.4.4 (now marked experimental again) is still considered the latest.
huh???
Example provided above (although other experimental packages might work) 1. "emerge search gtetrinet" initially shows: * app-games/gtetrinet Latest version available: 0.4.1-r3 Latest version installed: [something] Homepage: http://gtetrinet.sourceforge.net/ Description: Gtk Tetrinet Clone for linux If the 0.4.4 ebuild is manually installed, [something] is 0.4.4 and the "latest version availible" is still 0.4.1-r3 (although that's not really portage's fault). If gtetrinet is not installed, [something] is "not installed". 2. Edit /usr/portage/app-games/gtetrinet/gtetrinet-0.4.4.ebuild . Change the line 'KEYWORDS="~x86"' to read 'KEYWORDS="x86"' 3. "emerge search gtetrinet" now shows: * app-games/gtetrinet Latest version available: 0.4.4 Latest version installed: [something] Homepage: http://gtetrinet.sourceforge.net/ Description: Gtk Tetrinet Clone for linux 4. Edit /usr/portage/app-games/gtetrinet/gtetrinet-0.4.4.ebuild . Change the line 'KEYWORDS="x86"' back to 'KEYWORDS="~x86"', or resync to the current portage tree (as of today) to have the equivalent effect. This makes the package experimental again. 5. "emerge search gtetrinet" now shows: * app-games/gtetrinet Latest version available: 0.4.4 Latest version installed: [something] Homepage: http://gtetrinet.sourceforge.net/ Description: Gtk Tetrinet Clone for linux But the latest version marked non-experimental is 0.4.1-r3 again. Hence, emerge is reporting a funny result. ------------- I'm having a hard time finding more examples due to most people using portage.mask to hide their experimentals instead of 'KEYWORDS="~86"'; basically, it seems to be this: -- If a new version app is marked experimental via the "~86" KEYWORDS flag and is NOT being hidden by the package.mask master banfile, changing "~86" to "86" will make it be shown as the latest version. But changing the flag of the later (experimental) app version from "86" back to "~86' is _not_ enough to make the former non-experimental "latest entry" the latest entry again. This is really hard to explain because it is kind of obscure, and once I "corrupted" my system with the later entry, I have not been able to reset it to its former state. This likely will end up a WORKSFORME unless I've nailed it right on the nose.
*** Bug 10871 has been marked as a duplicate of this bug. ***
works correct here, so closing now.