It was also reported on bagtracker:
https://bugs.kde.org/show_bug.cgi?id=402835
https://bugs.kde.org/show_bug.cgi?id=429433

The other thing is:
This popup window can recognize if the site is already bookmarked or not.
In case it is already bookmarked it would be good to display keywords in keyword field.

According to documentation it should be closed by Qt.
https://doc.qt.io/qt-5/qfile.html#dtor.QFile

Thank you for your contribution.

• Use eventFilter for Back&Forward mouse buttons

• Revert "searchToolbar: Add placeholder for resultInfo in the ui file"

• searchToolbar: Add placeholder for resultInfo in the ui file
• Fix code style and remove unneeded code as requested

OK

That is what you think from user perspective.
But if there is ever any bug which triggers those functions without a break a 1000 times, users will be mad.

+ as drisca suggested, it would be great to add some stopper for notifications to show only once.
You may see it as "it will show only when there is a problem", well trust him with this, there will be some users who will be spammed with notifications.
I apologize I did not thought about this.

Can you return original blank lines ?
The spacing between lines is there to make it easier to read.

I am not that sure about it, you see I am more of a programmer than a designer,
iirc the messagebox will block Falkon until you press it so it is a bit annoying.
Maybe using falkon notifications sounds better.

I am against this patch.
In my eyes I expect it to just work and to not bother me.
I would welcome notification when it for some reason failed, because that would have some value for users.

In D26872#599904, @drosca wrote:

That's not true. If initialize fails then it doesn't make any sense to keep using the wallet as it is in "inconsistent" state.

In such case it should be totally rethought since the false can be triggered also by not having write permissions.
So i think we should in such case introduce new internal variable to check if we can write into Falkon folder.
Also at first run, we should probably always return true for migration thing, but need to make sure we are in Falkon folder.

Looks fine to me.

@drosca Even in destructor ?

To be more precise I think the condition should be in these methids

• addEntry
• updateEntry
• updateLastUsed
• removeEntry
• removeAll

Than you can add the condition into all methods.

Done in https://phabricator.kde.org/D26815

I would need your name and email which I can use to commit it for you.
(I can guess, but I want to be sure)
Thank you.

Do you have commit / push rights ?

Thank you for the fix.
Now it looks and works fine.

OK, works almost fine, only the checkbox disabled state is not triggered at the first launch of preferences dialog.

In D26753#598233, @alukichev wrote:

How about "(requires Pepper)" in the checkbox caption or a "This requires the option \"Allow Pepper Plugins (Flash)\" to be enabled" tooltip?

I know that.
I still believe we need to do something to make it more obvious or we would just push this work for later date.

So I tested it and it works as intended.
What I see as a problem is that this patch does not honour the relationship between PDF viewer and Pepper plugins.
So if you disable pepper plugins PDF viewer would also be gone and users would complain.

If there aren't any objections I will land this patch tonight.
It provides the same basic functionality as QtWebEngine default bubble.

it would, but I am not such great guru.
I tried few ways to change the behaviour but it was even worse,
and since this is used in many programs I do not want to break them.

This is the test I used to check the validity of my change, it kind of always fails.
I only hope someone can fix this widget at least for KF6.

• KMessageWidget: Add resize autotest

It looks good.
But I wrote a test as you suggested which still fails.
The test creates parent widget and puts the kmesagewidget and qlistwidget in it in vertical order (as in my test app)
and than it resizes the parent widget and compares the height (the available macros in *autotest.cpp)

No, my tests say no.
This does not update sizeHint which can cause some issue later.
It sure looks nice in my test app but it is wrong in the core.
This is not the right solution.

OK, I will try to write a test for it.
This change looks like it lacks some things to make it work properly in some cases.
I am not sure if I am the right person to fiddle with this code (but it is broken...)

So it is even more complicated with this PDF viewer than I thought.
Thank you for making the option.

It should be optional, many people would like to have internal PDF viewer.

It can be disabled, it is done as "plugin" in a similar way as flash player.
It can be disabled by unchecking simple checkbox at:
(Preferences > Browsing > [ ] Allow Pepper Pluginf (Flash player)).

• GreaseMonkey: Fix navigation type check for lower qt version than 5.14

Wait a bit, this works only for Qt 5.14.