> > I just saw that the i2c stuff from Greg's tree have been backed out > > from 2.6.4-rc1-mm1. Quoting the release notes: > > > > "bk-i2c.patch was dropped pending it getting its lmsensors act > > together." > > > > I don't quite understand what I am (well, what we are...) supposed > > to do (actually, I do not even get the meaning of the sentence > > itself). I clearly exposed my reasons for an interface change and as > > far as I know, nobody objected to that particular point. > > The problem is that there is no way a user can get a working > libsensors to work with the -mm kernel right now, correct? Not quite. The necessary patches are available on my personal server, as advertised here: http://marc.theaimsgroup.com/?l=linux-kernel&m=107787111318197 Maybe the address should have been advertised more. I was a bit surprised that the changes went into Andrew's tree for testing (although I now understand and agree that it was the right thing to do) and we suddenly had to support two incompatible trees with one single CVS repository. I thought we wanted to have our CVS repository always compatible with the latest Linus' tree, this is why I had not commited the sysfs interface change yet. And also because I did not want to "force" the kernel patch to be accepted. > I thought you were going to make a new release for the 2.6.3 kernel, and then > commit your changes to cvs to allow users running the -mm tree to use > that version. > > Is this the case? To the light of today's "incident", we should have been (remember I was in favor of a release from the beginning). Now that Linux 2.6.4-rc1 is out, releasing now is a bit odd (since we would more likely have to release again within a week or so). So the 2.8.5 release of lm_sensors will be for Linux 2.6.4. Linux 2.6.3 will be unsupported (only affects w83l785ts users). Anyway... Is it correct to provide the libsensors update as patches for mm kernel users (more publicly advertised, granted), or is it necessary to commit them to our CVS repository? I admit that I usually don't much like setting things up for temporary situations, but it looks like this is how you kernel folks expect things to be done... Let me know how you expect us to give people access to a mm compatible version of lm_sensors (basically: CVS or patches?) and we'll do juste that. Thanks and sorry for the trouble. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
