See bug 396025. If the dependency required is ">=linux-headers-2.6.19.2" then this should have been added and not ">=sys-kernel/linux-headers-2.6.39" which breaks every system that wants to have linux-headers < 2.6.39. Reproducible: Always
upgrade your kernel headers
(In reply to comment #0) > See bug 396025. If the dependency required is ">=linux-headers-2.6.19.2" then > this should have > been added and not ">=sys-kernel/linux-headers-2.6.39" which breaks every > system that wants to have linux-headers < 2.6.39. Andreas, any time we update a dependency version, that "breaks" every system wanting to use an older version. This is not a "break", but a consequence of using a rolling distribution. Mike, can you please provide one or two reasons why the user needs (should?) update the kernel headers? I believe the move to 2.6.39 headers was a "preventing" bump to avoid issues like the one in the related bug, but I don't have direct knowledge of this package.
if they don't update, then busybox doesn't build as described in the summary
The summary says ">=sys-apps/busybox-1.17.4 missing dependency on >=linux-headers-2.6.19.2 for mtd-user.h" and not ">=sys-apps/busybox-1.17.4 missing dependency on >=linux-headers-2.6.39 for mtd-user.h"
what exactly is your point ? upgrade your headers and then there's no problem.
I don't want to use kernels >= 2.6.39, mostly I use 2.6.32. You still did not answer the question why you introduced a dependency that is far to strict.
i didn't tell you to upgrade your kernel. i said upgrade your kernel *headers*.
Due to the nature of busybox you are likely to run into these types of problems from time to time. If you don't want to upgrade your headers then as a user you have the option of using 'savedconfig' which would let you disable any mtd related applets.
(In reply to comment #7) > i didn't tell you to upgrade your kernel. i said upgrade your kernel > *headers*. I understand that. http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=blob;f=Documentation/make/headers_install.txt says: > Kernel headers are backwards compatible, but not forwards compatible. > This means that a program built against a C library using older kernel > headers should run on a newer kernel (although it may not have access > to new features), but a program built against newer kernel headers may > not work on an older kernel. The text clearly says that you should NOT run older kernels (as I said I use 2.6.32) with newer kernel headers.
if you find such an app in the tree, file a bug
@userrel: Please take care of this case. Mike refused to answer the question why he insists on introducing this far to strict dependency. Official statements from the kernel developers did not convince him. I am building sys-apps/busybox-1.19.3-r1 successfully against older kernel-headers (2.6.32).
(In reply to comment #11) > @userrel: Please take care of this case. Mike refused to answer the question > why he insists on introducing this far to strict dependency. Official > statements from the kernel developers did not convince him. I am building > sys-apps/busybox-1.19.3-r1 successfully against older kernel-headers (2.6.32). IMHO Mike is the maintainer so he gets to pick what dependencies to put in (and support.) You are free to make an overlay for busybox so you can express whatever dependencies you prefer. This specific scenario (depending on a higher version than strictly necessary) is not uncommon in the tree. At some point we must decide a lower bound for the dep. You've tried convincing Mike. I talked to him in person at SCALE on Friday and he didn't seem to be interested in changing the dep. If it was something more complicated than making an overlay and changing the dep than perhaps I'd be more sympathetic to your case, but as it stands I think that is the best route forward. 2.6.32 is pretty old at this point (released Dec. 2009, it is Jan 2012, so over 2 years.) -A
