On Thu, Jun 12, 2014 at 04:57:01PM +0100, Ben Green wrote: > Quoting Corey Wright <undefined@pobox.com>: >> i see people hop on irc or post to the mailing list saying >> the latest official patch won't apply to the latest kernel >> version, but they haven't tried the patch i've provided (yes, >> it sometimes works across major kernel versions) nor offered >> to help herbert (technically or financially). >> i also understand that it's to herbert's benefit to care for >> the community so that people don't leave and he lose his >> testers and other benefits (but i'm also comfortable with him >> porting linux-vserver to newer major kernel versions and the >> community port to newer minor versions). >> anyways, just a suggestion to the community at large. > What I do with the repos is to only compile with the patches > Herbert releases. > They do take a while sometimes, but if there's any suggestion > you have as to how I could be part of speeding up that process > that would be great. Reporting back problems with good test cases is one way, providing patches to solve existing problems or improve missing virtualizations or add useful features is another. I'm happy to discuss any changes to the existing patchset as long as it doesn't mean extensive unpaid work for me. > Compiling an entire kernel to test a patch is a tall order > for most people, even those of us who do that sort of thing > regularly. Note that all patches I upload are compile and run tested, although it is impossible to test all options or filesystems for each and every patch. > I guess what I'm saying is that part "community port to newer minor > versions" sounds like it might have some mileage, but how? > How would we ensure sufficient testing of the community > patches? I think for minor changes a real hardware boot and guest start/stop test should suffice unless some specific sub- system (e.g. a filesystem, a network protocoll, etc) are affected (in this case a test of the subsystem would be required) > Is there a procedure we can follow that would mean Herbert > would be satisfied enough with the quality of the community > patches he'd allow them to distributed on the linux-vserver > website? We had and still have the grsec combos on linux-vserver.org, so I don't see a big problem with adding other patches from the community there as long as: 1) it doesn't mean more work for me i.e. it has to be automated in some way with clear responsibilities for those patches. 2) it is clear to the community that those patches are not 'official' because there is no way to guarantee that they are fine/safe without going through them line by line. Best, Herbert > Cheers, > Ben