On 2012-04-16 10:23 PM, Jeroen van Bemmel wrote: > Felix, > > Regarding the problem of users playing with config options and not > mentioning it, part of the solution could be to require users to submit > their .config files when reporting issues. In such a case, having the > kernel version in there too would actually be better than relying on > users mentioning the fact that they edited their Makefiles... I still fail to see how adding a config option for the kernel version makes a big difference in usability. I really don't consider the kernel version as something to be configured.
Also, I just found yet another reason why adding this is problematic: The menuconfig include file which provides the target selection has lines that select symbols like LINUX_3_2, which are used for various dependencies and checks throughout the system. Your patch does not handle that, hence it would introduce some subtle breakage if the kernel version is changed through menuconfig instead of a Makefile edit. I'm not interested in any further suggestions on how to slightly mitigate the issue of users creating bogus reports after having messed around in their configuration until I've heard some good reasons why this change is useful in the first place. The reasons I've heard so far are: - To help coordinate testing of new kernel versions => unnecessary, the people whose testing feedback we would actively consider for figuring out if a new kernel is ready are typically all able to do a simple edit in a Makefile. Typically testing a new kernel version only makes sense when the new version is actually ready for that, and developers typically actively ask people for testing when they need it during such a time. Outside of such a time, unsolicited testing feedback on newer kernel versions is usually ignored anyway, since there typically are some known issues to work out first. - To change the kernel version in device profiles => do not want!1!! (as explained earlier) - It helps people that are new to the system to play with stuff => It helps people find yet another way to mess up their builds without knowing what they're doing, we already have too many of those. Am I missing some compelling reason that I haven't already listed here and debunked? - Felix _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
