435885 - EntryCompletion popup not working properly

Status: RESOLVED FIXED

(Core :: Widget: Gtk, defect)

Description

[Antonio Gomes (tonikitoo)]

PRECONDITIONS: 
User have been use browser earlier (History is available).

STEPS LEADING TO PROBLEM: 
1. Open fennec.
2. Start writing url to the field => History list opened.
3. Continue write url

EXPECTED OUTCOME:
Url writing continues without any "breaks".

ACTUAL OUTCOME:
Every time when history list open, user needs to tap twice to character, which
he want to add.

FREQUENCY OF OCCURRENCE: 
Always. 
Reproducible with Fennec on Linux PC and Linux Arm (n800 and n810 devices)

Comment 1

[Mark Finkle (:mfinkle) (use needinfo?)]

I can't replicate this using the n810 keyboard. The action works as I would expect (no need to double enter a letter).

Comment 2

[Antonio Gomes (tonikitoo)]

(In reply to comment #1)
> I can't replicate this using the n810 keyboard. The action works as I would
> expect (no need to double enter a letter).
> 

mfinkle: you are right. Bug is reproducible only w/out physical keyboard, e.g. n800 with patches from bug 426344.

maybe we have a dependency here (?)

Comment 3

[Stuart Parmenter]

is this still an issue with m5?

Comment 4

[Antonio Gomes (tonikitoo)]

Attachment: patch v1

patch avoids entrycompletion to steal focus.

ps: this issue has been fixed in gtk side on maemo already by ifdef'ing some similar peace of code.

Comment 5

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I don't want to sprinkle #ifdef NS_OSSO through our code. I don't even know what OSSO is. Perhaps it would be better to have a global variable or a function so that this can be switched dynamically? With a better name, like gEnableVirtualKeyboard or something.

Comment 6

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

Roc, according to wiki.maemo.org "OSSO means "Open Source Software Operations". It's a department inside Nokia."  In our code, NS_OSSO has been used to designate the flavor of linux running on the n800/n810.  Perhaps NS_MAEMO would be more appropriate...

Antonio, is this a problem specific to hildon?  If so, we should be using NS_HILDON.

Comment 7

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I honestly don't think we should use NS_MAEMO or NS_HILDON either, although we already have used it. I'd prefer to use names that have some actual technical meaning, i.e., something describing the thing that's different about Maemo/Hildon that means we have to behave differently there.

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

OK, if "Hildon" is the official name for a variant of the GTK2 platform, then I suggest we use MOZ_WIDGET_GTK2_HILDON.

Comment 9

[Stuart Parmenter]

i'm not very concerned about the ifdefs, however, I'm concerned about us changing basic functionality based on one without a really big comment above explaining why it is OK.  Should it only be this way when only when the virtual keyboard is up?  Why is it OK to always be that way?  What is special about Maemo/Hildon?

Comment 10

[Antonio Gomes (tonikitoo)]

Attachment: patch v2

patch on top of already r'ed+ patch in bug 406837.

added 'gIMEVirtualKeyboardOpened' to avoid EnsureGrabs() to get call only when vkb is up.

Comment 11

[Antonio Gomes (tonikitoo)]

gtk maemo guys did something similar for their gtk+-2.10 changes

see http://mxr.maemo.org/diablo/source/gtk+2.0-2.10.12/gtk/gtkentrycompletion.c#1592

Comment 12

[Antonio Gomes (tonikitoo)]

Attachment: v3 - ready to go. [Checkin: Comment 13]

same patch but changed NS_HILDON by NS_PLATFORM_HILDON ... see bug 453867.

Comment 13

[Serge Gautherie (:sgautherie)]

Comment on attachment 338155 [details] [diff] [review]
v3 - ready to go.
[Checkin: Comment 13]

http://hg.mozilla.org/mozilla-central/rev/31181a76dd29