Nelson Castillo wrote: > I'm sending a patch that does both things. Thanks !
Mixing moves with small changes always leaves some uneasiness. Actually moves generally do, whether there are changes or not. This is because nobody is going to verify if you really just moved the file or if you also changed it, possibly by accident. It's a bit of a pity that, while you seem to use git as well, the submissions from you to Andy are still by mailing patches instead of directly pulling from your repository. Otherwise, you could just have posted a git-diff -M for review purposes, and all those moves would have ended up just as rename from include/linux/ts_filter_linear.h rename to drivers/input/touchscreen/ts_filter_linear.h etc., and it would have been obvious whether and where you had changed something else. Unfortunately, there doesn't seem to be a way to get git to use renames in the summary but a regular diff in the actual changes. Maybe it would make sense to post a version with -M and one without in such cases. > Well, nothing from linux-input. In LKML Andrew Morton just told us to > fix a few things. I think it is a good thing that nobody has told us > that we suck, not yet :-) It's not a bad start :-) > We need to submit code that actually uses the filters before they > accept the patches upstream. Right now s3c2410_ts.c and mach-gta02.c > come to my mind (I am not aware of the dependencies). mach-gta02.c is one of our most hellish files for sure ;-) I think it would be worth breaking it up into small parts by subsystem, and move the whole rabble in its own directory. This is without precedent in the ARM tree so far, but then nobody else has quite as much stuff there as we do. The closes thing so far would be mach-at91/at91sam* which has a mixed approach using two files. Might be worth getting the opinions of linux-arm-kernel on that. - Werner
