989281 - Auto-adjust brightness is too bright when in darkness

Status: RESOLVED WONTFIX

(Firefox OS Graveyard :: Gaia::System, defect, P2)

Description

[Frederik Braun [:freddy]]

I know it's hard to calibrate this and I know there is bug 809645 about the screen being to *dark* :)
But when in utter darkness, the screen is horribly bright, which makes it really uncomfortable to use the phone in "auto adjust" brightness.

I am experiencing this with the Alcatel One Touch Fire (Hamachi?) and the Flame device.

Comment 1

[Julien Wajsberg [:julienw]]

Do you still see this?

Comment 2

[Frederik Braun [:freddy]]

Tricky question, I did not take measurements, when I filed this in March ;-)

I just tried and it looks quite OK. I mean, it's probably not a great idea to look at a small glowing screen in utter darkness anyway...

Did someone touch the brightness since then? If so, I think it's OK to close this as worksforme/duplicate.

Comment 3

[Julien Wajsberg [:julienw]]

I'm not sure, for this specific issue :)

I know that flame-kk's sensor driver returns lower values than flame-jb, and as a result the resulting brightness that we set is lower. But in real darkness it should really be the same: 0.

To be really precise, when there is no light at all, the ambient sensor does not send any event, and as a result I'm not really sure how we can set something. Maybe that is the issue you saw here?

Comment 4

[Julien Wajsberg [:julienw]]

The more I think of it, the more I think this is the real underlying issue.

Then that likely makes it a vendor issue, but I don't know how to move forward.

Comment 6

[Julien Wajsberg [:julienw]]

I have a workaround in mind: when enabling autobrightness, if we got no value from sensor in ~1sec, we can assume it's because we're in dark condition and set the lowest brightness.

Comment 7

[Julien Wajsberg [:julienw]]

I'll try this.

Comment 8

[GitHub Autolander Bot]

Attachment: [PullReq] julienw:989281-set-low-brightness-in-low-light-conditions to mozilla-b2g:master

Comment 9

[Julien Wajsberg [:julienw]]

I'm not sure it improves much yet, I haven't tested this in dark conditions yet.

Moreover the underlying issue could still bite us when going quickly from a bright to a dark condition (because, again, dark conditions do not yield a sensor events... although we need to verify if this is the case when transitioning).

Comment 10

[Julien Wajsberg [:julienw]]

I also need to check if the assumption from comment 6 is valid. For example, maybe we don't get a sensor event if the value just does not change (so if we stay at constant ambient light)? I'll need to check on other devices like Buri.

Comment 11

[Julien Wajsberg [:julienw]]

(In reply to Julien Wajsberg [:julienw] from comment #10)
> I also need to check if the assumption from comment 6 is valid. For example,
> maybe we don't get a sensor event if the value just does not change (so if
> we stay at constant ambient light)? 

Unfortunately, it seems that's what happens :(

Here is what happens with the patch:

1. disable the autobrightness option, and set screen at max brightness
2. enable it, try to stay within a same light environment
=> with my patch the brightness drops to the minimum brightness, supposedly because the light does not change and thus we don't get an event.
3. as soon as the ambient light changes and we get an event, the screen gets an appropriate brightness.


It think this is better than the current situation, but still not ideal.

Short term, the best would be to do what Jonas says in bug 1057185 comment 12...

NI Jonas and Tim to try to get some feedback.

Comment 12

[Tim Guan-tin Chien [:timdream] (please needinfo)]

bug 1057185 comment 12 might not be the best idea as an event always means change in value and to get the current value we should always look at a property. I don't understand what stopping us if we only expose such property to certified app.

I would recommend we stop working on this bug because any fix we got here will probably be device dependent...

Comment 13

[Julien Wajsberg [:julienw]]

What would be device-dependent? The current patch is not device dependent, but assume something that's not true in all situations. It's more situation-dependent :)

I think the situation is better with the patch that without the patch, because it gives a feedback to the user: when the user sets the "auto-adjust" setting, the screen brightness will always change.

The issue I'm outlining in comment 11 is really uncommon in my opinion.

That said I couldn't test on another device than Flame; on my Buri the ambient light sensor looks broken.

Comment 14

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Tim: Is your concern that current code assumes that "an event was fired" means "there was a change in value"? So that for example code that listens to the proximity sensor will assume that every other event means "close" and every other event means "not close" rather than looking at event.value?

Do you know of existing code that makes such an assumption or are you "just" concerned that there might be such code?

It seems like a pretty bogus assumption of code to make since in theory sensors could change the value from "close" to "pretty close". I.e. from 0 to 0.1 rather than from 0 to 1.

That said, code might definitely still make such bogus assumptions and fall over if we break them.

Or do you have other concerns?

Comment 15

[Julien Wajsberg [:julienw]]

Auto-adjust brightness logic is in Gaia::System.

Comment 16

[BMO Automation]

Firefox OS is not being worked on