Created attachment 302596 [details] [review]
wacom: Remove calibration notification

On sane hardware it is mainly unneeded, and the UI on g-c-c is
discoverable enough for the cases where it isn't, so let's avoid
nagging about something mostly superfluous.

Review of attachment 302596 [details] [review]:

Sure. Should be committed to 3.16 (possibly 3.14?)

Attachment 302596 [details] pushed as 015f050 - wacom: Remove calibration notification

It's a bit late with the patch now pushed upstream, but I was wondering if Carlos wouldn't mind going into a bit more detail about the IRC discussion, with particular regard to the reasons that it was decided to remove the notifications entirely.

Based on what's written above, it sounds like notifications were removed because they were more annoying than useful. Although I would agree this was the case, I feel that they wouldn't be annoying if they only appeared when actually necessary. Even ignoring bugs which caused the notification to appear on every boot (#741738), the notification would appear any time the resolution changed, which was a poor signal to determine calibration need.

Calibration is done primarily to counteract parallax effects. These don't necessarily change when the resolution does, but will change significantly when the device is rotated. If g-s-d watched for rotation signals and cached the calibration results for each orientation (bug #737758) then the user would only see the notification when it might prove beneficial.

IMHO the notifications are valuable when used appropriately and could have been fixed rather than removed entirely. Were there additional arguments against that were discussed?

(In reply to Jason Gerecke from comment #4)
> It's a bit late with the patch now pushed upstream, but I was wondering if
> Carlos wouldn't mind going into a bit more detail about the IRC discussion,
> with particular regard to the reasons that it was decided to remove the
> notifications entirely.

Sure, let me paste the IRC conversation for reference (#control-center from Apr 28th):

bentiss: hadess, garnacho_: ping -- hey guys, I have some complains regarding the wacom plugin :)
garnacho_: bentiss: hey :)
bentiss: so cshalle asked me to look into an issue with the Toshiba Z20t
bentiss: this laptop has an internal Wacom display with no stylus shipped in the box
bentiss: so each time you log in, change VT, or plug a monitor, you get a notification asking you to calibrate the screen but you can not do it :/
bentiss: garnacho_: that's the first problem, more to come :-P
* garnacho_ all ears :)
bentiss: then I tried F22 on the Cintiqs we have, and it looks like the Cintiqs are also showing up the calibration popup
bentiss: while I wouldn't expect anybody to calibrate a cintiq at all
bentiss: second problem
hadess: the cintiq is pretty much the only thing you'd want to calibrate...
bentiss: no
hadess: at least that's why we've implemented the feature at all
bentiss: it works really well out of the box, and I don't think the calibration tool can fix the small problems that can appen at the border of the cintiq (the calibration would need to be non linear)
hadess: then i'm not sure why we have that feature at all
bentiss: and when the screen is rotated, remapped and everything, we should be doing like we do for touchscreen, handle this transparently
hadess: how did you rotate the screen?
garnacho_: calibrating would still sound useful on odd resolutions I guess
bentiss: I don't but that's one of the requirements for the calibration tool when you have an internal ISD wacom (on a tablet PC) -> that's all IIRC
garnacho_: not sure how often will designers set displays at lower resolutions though
mclasen: of course, we can never remove it, lest the crowds scream 'gnome is removing features again!'
bentiss: true, but we should not pop it up every time, but keep a button in the wacom UI IMO
bentiss: ok, now to the third problem (I opened a gnome BZ this time): https://bugzilla.gnome.org/show_bug.cgi?id=748589
Services: Bug 748589: normal, Normal, ---, carlosg, NEW , Screen mapping incorrect for Cintiqs
bentiss: looks like on my setup the Cintiqs are mapped to the internal scren of my laptop
bentiss: garnacho_: it might be a problem from my side, so if you could just give a quick test and reproduce or close it that would be great
garnacho_: bentiss: are maybe the 2 displays too similar in size?
garnacho_: bentiss: plugging the cintiq now :)
bentiss: garnacho_: I tried on the t540p with the cintiq 22Hd and the 21UX
bentiss: so 15" vs 21/22"
garnacho_: aha, definitely not then
garnacho_: unless edid lies, but that'd be weird on wacoms I guess
bentiss: the resolutions might be close though (the t540 is a 1080p display IIRC)\
bentiss: OK, dropping my last complain while you are testing: the screen calibration tool is really hard to be removed. There is no indications that you can just press Esc or alt-F4, and it crashes in some circumstances (Jason filed #747880)
bentiss: hadess, garnacho_: sorry for raising so many problems, but Wacom is somewhat important for Red Hat (and gnome)
bentiss: garnacho_: if you need me to file/point at bugs, say so
garnacho_: bentiss: hmm, I think it would be great to get --debug g-s-d logs for the misassigned output
bentiss: garnacho_: OK, I'll add the info on Thursday when I will be in the office
garnacho_: bentiss: thanks :), I can indeed reproduce jason's bug in turn
garnacho_: bentiss: as for the other issues I guess it comes down to "don't show the notification as pervasively"
mclasen: or ever ?
mclasen: if we conclude that calibration should not be needed, why notify ?
bentiss: mclasen, garnacho_: I would agree to that, yes :)
garnacho_: mclasen: yeah I guess... my only usecase with this in mind is say 4:3 resolutions on 16:10 monitors, and I doubt anybody will do that on purpose
bentiss: garnacho_: even for that, you should be able to remap it directly in gsd, you know the size of the displays, so applying the xinput/libinput matrix should not be a problem
garnacho_: indeed
garnacho_: bentiss: I'll make a patch killing the notification and put it on a bug
bentiss: garnacho_: many thanks
bentiss: garnacho_: though we must first checks that the internal gsd mapping works first (or people will complain)
bentiss: I can also test that on Thu.
bentiss: well, once my setup works for the cintiq :)

> 
> Based on what's written above, it sounds like notifications were removed
> because they were more annoying than useful. Although I would agree this was
> the case, I feel that they wouldn't be annoying if they only appeared when
> actually necessary. Even ignoring bugs which caused the notification to
> appear on every boot (#741738), the notification would appear any time the
> resolution changed, which was a poor signal to determine calibration need.
> 
> Calibration is done primarily to counteract parallax effects. These don't
> necessarily change when the resolution does, but will change significantly
> when the device is rotated. If g-s-d watched for rotation signals and cached
> the calibration results for each orientation (bug #737758) then the user
> would only see the notification when it might prove beneficial.

I agree that might have been another way to go around this. In the end, I think it'd be great to have g-s-d conveniently ignore the unused bands of the display in lower resolutions, and apply these on the right sides when rotating.

> 
> IMHO the notifications are valuable when used appropriately and could have
> been fixed rather than removed entirely. Were there additional arguments
> against that were discussed?

None other than the stuff discussed in IRC, as pasted above. Recently before that I reinstalled stuff on an N-trig powered laptop, and the first thing greeting me was this notification, on the laptop natural resolution/orientation, so I might have been biased.

*** Bug 737758 has been marked as a duplicate of this bug. ***

*** Bug 732600 has been marked as a duplicate of this bug. ***

*** Bug 735981 has been marked as a duplicate of this bug. ***

*** Bug 741738 has been marked as a duplicate of this bug. ***