714950 - Unable to see characters typed in landscape fullscreen VKB

Status: VERIFIED FIXED

(Firefox for Android Graveyard :: General, defect, P2)

Description

[Aaron Train [:aaronmt]]

On the bug URL: http://mail.yahoo.com, rotating device to landscape any input I enter into the fields 'Yahoo! ID' or 'Password' is not displayed in the edit-box of the virtual-keyboard. This looks like a regression.

Comment 1

[Aaron Train [:aaronmt]]

Tested on:
Motorola Droid Pro (Android 2.3)
Mozilla/5.0 (Android; Linux armv7l; rv:12.0a1) Gecko/20120103 Firefox/12.0a1 Fennec/12.0a1

Comment 2

[Alex Pakhotin (:alexp)]

This is a more general issue. The problem is with native fullscreen keyboard UI. As I already mentioned in the bug 708774 it was broken by the recent major change to the GeckoInputConnection class. The input handling has been significantly simplified to fix the issues we had, but that caused some side-effects.
I am working on this.

Comment 3

[Alex Pakhotin (:alexp)]

(In reply to Alex Pakhotin (:alexp) from comment #2)
> it was broken by the recent major change to the GeckoInputConnection class.

For the record - that was bug 595008, so it's a known point of regression.

Comment 4

[Tony Chung [:tchung]]

Attachment: screenshot on Galaxy Nexus

happens on Galaxy Nexus, both Aurora and Nightly.  2012-01-04 build, 11.0a2 and 12.0a1

Repro'd on google.com

Comment 5

[Tony Chung [:tchung]]

Attachment: screenshot on SGS2

surprisingly, i can't reproduce on my SGS2, android 2.3.3

Comment 6

[Alex Pakhotin (:alexp)]

(In reply to Tony Chung [:tchung] from comment #5)
> Created attachment 585973 [details]
> screenshot on SGS2
> 
> surprisingly, i can't reproduce on my SGS2, android 2.3.3

This is not a fullscreen VKB. The problem we have is only with so called "extracted text UI" - when IME opens its own UI with an edit box, which covers the application completely.

Comment 8

[Alex Pakhotin (:alexp)]

Attachment: Fix

Update extracted text UI on the content changes.

The patch also contains some debug output tweaking for GeckoInputConnection.

Comment 9

[Brad Lassey [:blassey] (use needinfo?)]

Comment on attachment 586631 [details] [diff] [review]
Fix

Review of attachment 586631 [details] [diff] [review]:
-----------------------------------------------------------------

::: mobile/android/base/GeckoInputConnection.java
@@ +210,5 @@
> +
> +        return setSelectionInternal(start, end);
> +    }
> +
> +    private boolean setSelectionInternal(int start, int end) {

why do we need setSelectionInternal()? just call super.setSelection() in setSelection();

Comment 10

[Alex Pakhotin (:alexp)]

(In reply to Brad Lassey [:blassey] from comment #9)

> why do we need setSelectionInternal()? just call super.setSelection() in
> setSelection();

It is also called from notifySelectionChange(). I wanted to add a method, which does stuff specific to this subclass, but without calling Gecko. Right now it simply calls the super class. Of course could call super.setSelection() directly in both places - I wasn't quite sure what would be better, an extra layer just seemed a bit safer for the future.

Comment 11

[Alex Pakhotin (:alexp)]

Attachment: Fix v2

Got rid of setSelectionInternal().

Comment 12

[Brad Lassey [:blassey] (use needinfo?)]

Comment on attachment 587572 [details] [diff] [review]
Fix v2

Review of attachment 587572 [details] [diff] [review]:
-----------------------------------------------------------------

::: mobile/android/base/GeckoInputConnection.java
@@ +629,3 @@
>          switch (type) {
>          case NOTIFY_IME_RESETINPUTSTATE:
> +            if (DEBUG) Log.d(LOGTAG, ". . . notifyIME: reset");

why change this logging?

@@ +650,5 @@
>              IMEStateUpdater.enableIME();
>              break;
>  
>          case NOTIFY_IME_CANCELCOMPOSITION:
> +            if (DEBUG) Log.d(LOGTAG, ". . . notifyIME: cancel");

why change this logging?

@@ +655,5 @@
>              IMEStateUpdater.resetIME();
>              break;
>  
>          case NOTIFY_IME_FOCUSCHANGE:
> +            if (DEBUG) Log.d(LOGTAG, ". . . notifyIME: focus");

why change this logging?

Comment 13

[Alex Pakhotin (:alexp)]

(In reply to Brad Lassey [:blassey] from comment #12)
> >          switch (type) {
> >          case NOTIFY_IME_RESETINPUTSTATE:
> > +            if (DEBUG) Log.d(LOGTAG, ". . . notifyIME: reset");

It's just for consistency with other similar logging in this class, and better readability of the debug log.

Comment 14

[Alex Pakhotin (:alexp)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/939f9467fc73

Comment 15

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

https://hg.mozilla.org/mozilla-central/rev/939f9467fc73

Comment 17

[Martijn Wargers (dead)]

Verified fixed in today's build.

Comment 18

[Brad Lassey [:blassey] (use needinfo?)]

Comment on attachment 587572 [details] [diff] [review]
Fix v2

[Approval Request Comment]
Regression caused by (bug #): 
User impact if declined: 
typed chars can't be seen in landscape mode
Testing completed (on m-c, etc.): 
Risk to taking this patch (and alternatives if risky):
regressions for untested IMEs

Comment 19

[Alex Keybl [:akeybl]]

Comment on attachment 587572 [details] [diff] [review]
Fix v2

[Triage Comment]
Mobile only - approved for Aurora.

Comment 20

[Brad Lassey [:blassey] (use needinfo?)]

https://hg.mozilla.org/releases/mozilla-aurora/rev/b6ad62e6f637