On Wed, 2020-11-25 at 14:41 -0500, Mark Pearson wrote: > On 25/11/2020 11:42, Hans de Goede wrote: > > Hi, > > > > On 11/24/20 4:30 PM, Bastien Nocera wrote: > > > On Sat, 2020-11-21 at 15:27 +0100, Hans de Goede wrote: > > > > Hi, > > > > On 11/20/20 8:50 PM, Barnabás Pőcze wrote: > > > > > Hi > > > > > 2020. november 15., vasárnap 1:44 keltezéssel, Mark Pearson > > > > > írta: > > > > > > > > > Bastien, what do you think about Barnabás' suggestion to always > > > > have the files present and use poll (POLL_PRI) on it to see > > > > when > > > > it changes, listing maybe "none" as available profiles when > > > > there > > > > is no provider? > > > > > > Whether the file exists or doesn't, we have ways to monitor it > > > appearing or disappearing. I can monitor the directory with > > > inotify > > > to see the file appearing or disappearing, or I can monitor the > > > file with inotify to see whether it changes. > > > > Ok, do you have a preference either way? I personally think having > > the file only appear if its functional is a bit cleaner. So if it > > does not matter much for userspace either way then I suggest we go > > that route. > > > My (limited) vote is having the file appear when the profile is > loaded > seems nicer (and to disappear if the profile is unloaded). No > profile, > no interface just seems elegant. > > I pretty much have v4 of the patches ready to go (I wanted to rebase > on > the update for the palm sensor and finished that today). I'm happy to > hold off on them until we're ready if that helps. Bastien - let me > know > if it would be useful to have a version to test against to see what > will > work best for you, or if you need something different. Definitely > want > to make sure user space gets the best option as my understanding is > changing this later becomes a pain :) I don't have the hardware. I agree with you that "no profile = no file" seems cleaner, but whichever is easier for you.
