| Summary: | net-firewall/iptables-1.4.6 build failure with sys-kernel/linux-headers-2.6.33 | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Jack Lloyd <lloyd> |
| Component: | Current packages | Assignee: | Gentoo's Team for Core System packages <base-system> |
| Status: | RESOLVED INVALID | ||
| Severity: | normal | CC: | pva |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | iptables-1.4.6 build log | ||
|
Description
Jack Lloyd
2010-06-09 14:33:09 UTC
Hello JAck, can you please attach the complete build.log? Thanks Created attachment 234731 [details]
iptables-1.4.6 build log
Build log attached.
BTW, #s on the headers:
2.6.30:
ls /usr/include/linux/netfilter_ipv4 | wc -l
46
2.6.33:
ls /usr/include/linux/netfilter_ipv4 | wc -l
14
not a bug in linux-headers. this is what upstream linux wants. It looks like this was fixed upstream in January: https://git.netfilter.org/cgi-bin/gitweb.cgi?p=iptables.git;a=commit;h=350661a6eb089f3e54e67e022db9e16ea280499f So 1.4.7 and 1.4.8 should be OK, though I haven't tested. Perhaps a version mask is all that is required here. and 1.4.7 is in the tree already (for months at this point) How is this invalid? 1.4.6 still doesn't build you're breaking the system by mixing and matching stable and unstable And setting a DEPEND for !sys-kernel/linux-headers>=2.6.33 is that hard? we track two sets of trees synced to each other. issues like this dont come up when things are tracked correctly. by the time the kernel headers stabilize, iptables would stabilize as well. so yes, it is a waste of time because fixing one of these encourages people to find & file more. vapier: Having stable iptables NOT compile on a system that's otherwise ~arch system would be a good reason to have that blocker added. no, it isnt Mike, I don't think it's hard to add such code to make people happy, so I've added it. But in general I agree this bug is INVALID: it's impossible to track stable/unstable matching and with this fix I'm not going event to pretend that something was improved in the tree. New packages enter ~arch on daily basis and similar breakages exist all over the tree and it's completely impossible to track them. That said, if users provide solution - I tend to apply it. Also note this fix will die very soon since I'm going to fill stabilization request for 1.4.8 as soon as time comes... i'm not inclined to even let a trickle in so as to avoid encouraging more. think this is a good idea ? then i look forward to the people running stable packages with unstable gcc and filing bugs about those too. it's a terrible idea and the reason we have trackers/stabilization bugs in the first place. but i'm not going to take it so far as to revert your changes to iptables if you were so inclined to spend the time to make it. BTW, it might be useful to update the documentation to mention that mixing stable and unstable 'breaks the system'. http://www.gentoo.org/doc/en/handbook/handbook-x86.xml?part=3&chap=3 for instance simply describes how to do it with zero caveats. Jack, open a new bug for our documentation team to consider. I'm not sure that they'll add this since there is cation about ~arch in handbook, but IMO separate sentence about mixing could be given too. In any case this bug is not a place for such requests since we are not working on documentation. thanks. |
