I think the following dialogue speaks for itself: root@Delacroix /var/tmp/portage/libpng-1.2.5-r6 # qpkg -I libpng -v media-libs/libpng-1.2.5-r4 * root@Delacroix /var/tmp/portage/libpng-1.2.5-r6 # emerge =media-libs/libpng-1.2.5-r4 Calculating dependencies emerge: there are no masked or unmasked ebuilds to satisfy "=media-libs/libpng-1.2.5-r4". !!! Error calculating dependencies. Please correct.
the old versions got removed due to denial of service vulnerability, bug 49887 latest is media-libs/libpng-1.2.5-r6 and has been marked stable on all arches i don't consider this a bug
I apologize. I didn't know of any other place to put this. BTW, shouldn't they be hard-masked and not removed outright?
As far as I know, removing an ebuild is like hard-masking it. It can't be emerged anymore and portage will automatically update/downgrade to the latest version available.
If I come across something that's hard-masked, I ususally suspect that those have some sort of security flaw, and if I'm curious enough, I'll search the tracker/forums for answers as to why it was masked. If a few ebuilds suddenly disappear, I suspect either a portage malfunction, an incompetent maintainer, or a hacker in the portage tree, most likely the first. Functionally, as far as I know, you're right. But masking vs. outright removing looks cleaner and is less likely to cause concern/hysteria/unnecessary bug reports. Perhaps we should move this discussion to the forum?
Just discovered another reason to keep bad ebuilds in the portage tree, masked (and this looks rather serious) A tool I was experimenting with spat out this exact error(specifically, ecatmur's pruneworld, I called the tool with arguments -aP). It looks rather serious: !!! The package !!! media-video/nvidia-kernel-1.0.5336-r1 is !!! installed, but does not have a corresponding ebuild in an accessible !!! portage tree. Things will break. Correct this. NOW.
umm that's the point of keeping the ebuild in /var/db/pkg/ ... if it gets removed from the portage tree, it's not a problem in other words, comments 0-4 have no meaning ... this bug covers comment 5
That error is not a portage message.
