• Change normal bg to 15% opacity
• Change minimized bg to 5% opacity

Update patch according to discussion.
extractManagedFilePath will be broken in case of long truncated path but that has always been the case...

In D25360#573798, @IlyaBizyaev wrote:

Should this also affect GTK apps? There's a matching radio group in the GTK KCM.

Should this also affect GTK apps? There's a matching radio group in the GTK KCM.

In D25340#573672, @ndavis wrote:

@niccolove Do you mind if I commandeer this? I know dealing with these kinds of issues can be super annoying and I'm already used to it.

• For real

• Adapt deprecation version

Rebase

In D25699#573713, @fvogt wrote:

The check for the prefix was added in bf1d1cc6b2ad37cb586f44b56fa2438ed3a5dbfc, while the control.flat one got added much earlier.

The labels are visible again with just the control.flat condition, but the prefix one might be needed as well for non-breeze themes.

There's a part missing though, the triangle (visible on https://openqa.opensuse.org/tests/1105226#step/start_wayland_plasma5/21) is gone. That seems to be a feature lost with PC3 :-(

The check for the prefix was added in bf1d1cc6b2ad37cb586f44b56fa2438ed3a5dbfc, while the control.flat one got added much earlier.

Assuming we install all parts under plugins/parts instead of just plugins/ like we currently do, the application should be able to do

@niccolove Do you mind if I commandeer this? I know dealing with these kinds of issues can be super annoying and I'm already used to it.

ping!

It's going to be tedious with manual editing. Maybe there's a way to automate it with awk, but I don't know that tool well enough. This should at least make the deleting part easier: sed --follow-symlinks -i 's/style="opacity:0.05"//' tasks.svg

Oh wait, here's the problem.

In D25340#573654, @ngraham wrote:

Hmm, I'm not seeing any change, and the difference between open and maximized items is still too subtle IMO:

If the background is a different color/opacity, I can't tell.

Hmm, I'm not seeing any change, and the difference between open and maximized items is still too subtle IMO:

EDIT: I should probably format the added code with clang-format, I'll do that soon...

Reset tabbar to master

In D25340#573594, @ndavis wrote:

Wait, is a change to tabbar.svg supposed to be in this diff?

Wait, is a change to tabbar.svg supposed to be in this diff?

In D25334#566992, @ngraham wrote:

In D25334#566745, @ndavis wrote:

Yes that's right, the Kickoff tabs are fine, it's the panel that doesn't look right. It just doesn't look good with that kind of highlight to me. Or maybe the highlight background is too bright.

To me it looks fine with Breeze light IMO. If you think it looks bad with Breeze Dark, that seems like kind of a problem since as far as I was aware, this is the proposed new highlight style that you're working on, right? It definitely needs to look good with Breeze Dark, or at least no worse than the status quo.

Changed inactive to 5%

How would i load a "KPart that can open PDF files" then?

Yes it works :)

Address comments

Rebased, compiled and tested. Changes work as expected.
Double checked file_unix.cpp. Would be a shame if it happens yet again.

In D23384#571455, @sitter wrote:

The opposite extreme is to always pass when X-KDE-Protocols is set and assume that the applications are actually working correctly (e.g. vlc ought to talk to kiod/KPasswdServerClient to get credentials, otherwise its declaration of X-KDE-Protocols is incorrect and you are looking at a bug in vlc. at the very least it should throw up its own auth dialog if it doesn't know what to do).

With regards to the Docker/Gitlab CI part, please use the images under kdeorg/ on Dockerhub rather than personally maintained images as the wider community has no access to your namespace on Gitlab.com

Feedback on the API question would be welcome.

I guess you were expecting a higher-level review, but I don't know anything about these protocols.

Oh sorry, I missed that. It's the kind of information I wouldn't expect in the commit message.

Yep, in fact I mentioned this in the description section of the patch:

Please commit it after the commit for the new Frameworks is made (so probably from Sunday onwards).

I'm concerned that you didn't compile this (because of dependency issues, from what I gather), which means it's not tested either.

I guess I should change my status to accepted given that I think this is good enough and already an improvement. But I think we can do even better, @niccolove. :)

I see. This answers my question about why two merge requests -- no problem, keep them separate, but commit the fix before the unittest
[this is so that bisecting never ends up in the situation where unittests are broken]

In D25340#563400, @ndavis wrote:

This diff is against commit 467d721cc96258b54048c0dd1508d16e03c0cd55, which isn't in git master. Do I actually need that commit for this patch to work?

Welp, there's nothing objectively wrong with making this patch. LGTM

There was a bug report about it that had some people agreeing with it. I think it makes a bit of sense because yes, this dialog is indeed only about keyboard shortcuts, and at least to my ears, the phrase "keyboard shortcuts" instantly connotes what this is about, while "shortcuts" is a more generic term that doesn't have such an instantly recognizable computer-related meaning.

This doesn't seem wrong, but why is it needed? Do people get confused about the type of shortcuts? Are there non-keyboard shortcuts? If there are, would we put their configuration menu under a different menu option?

Can we push this forward? I just triaged a bunch of bugs and found that https://bugs.kde.org/show_bug.cgi?id=370465 now has five duplicates. There seems to be quite a bit of demand for this.

Oh no don't misunderstand me. I am very glad that someone else stepped up! I have more than enough work on my plate with Kdenlive so please help :)
Hopefully David can give us some hints about the best way to move on!

Maybe instead of "Unavailable", it could say "Not Running"?

Yeah, let's keep this the way it is. I think it's plenty comprehensible, and we never managed to come up with a good place to move them anyway.

Addressed the issues.
Annotated code to be removed in KF6.
My build env is outdated so I couldn't compile the new depracation macro. Copied as it is.

In D24489#573152, @mardelle wrote:

The unittest is in another diff because it's a different author I guess.

[...]
I didn't mean to step on your toes; feel free to disregard the unit test I wrote, or reuse whatever part(s) of it as you see fit. It's just that I was interested in the matter and sort of wanted to see if my understanding was correct. :)

In D25296#563291, @ngraham wrote:

Never mind, I wasn't deleting the cache files properly after rebuilding. When I do that, the monochrome icons don't get used at all and it reverts to the colorful one:

The two links I know of are:
http://0pointer.de/lennart/projects/libcanberra (works as is in file but
wont work as https)
http://www.x86-64.org/documentation/abi.pdf (http://www.x86-64.org is dead)

@ndavis or @trickyricky26, could I ping you on this?

• Fix indentation

The unittest is in another diff because it's a different author I guess. I can process the comments but looking closer I found 2 other important issues in this class:

In D25753#572894, @winterz wrote:

please send me a list of urls that don't have https: and I'll add them to the whitelist

In D25682#570849, @sitter wrote:

In D25682#570845, @davidedmundson wrote:

Why do we need to mirror this dsoap-ws-discovery-client lib that seems to be copied from somewhere?

Testing convenience until @caspermeijn makes a release mostly. The library was,or maybe even still is, in flux as API was being shuffled up to kdsoap.

Doesn't seem relevant anymore

Alternative approach without Q_NAMESPACE_EXPORT, which is too new

Haha yes almost exactly like that! Maybe today I'd use a sheet instead of a separate dialog window. But I have great faith that the 2019 David is able to conquer the challenges of 2009. :)

That is excellent!

That is excellent! But in the implementation, mind the "multiple kxmlgui-clients" use case, like konqueror or kdevelop (or kate I guess). It can make things tricky...
(you drop between two clients, how do you know which one should memorize the action into its kxmlgui file?)

Trying to assign anything to Shift+Delete shows the "conflicting shortcuts" error dialog.

is something you can use if you feel bored, i made it mandatory in poppler

@ngraham
Like this: http://static.davidedmundson.co.uk/toolbars_change_01.ogv ?

Oops, sorry! I added comments to D8773.

In D8773#167625, @colomar wrote:

In D8773#167607, @broulik wrote:

What I would like to see is that when you open the "Edit Toolbar" dialog that you then can drag around tool bar buttons and drag them from and to the dialog, like Firefox does it and how @colomar suggested. Given you can already re-arrange toolbar buttons whilst said dialog is open, what needs to be added is the ability to drag those from and to the dialog.

Yes, exactly. The closer to Firefox' "Customize" UX, the better. No more separate dialogs, just one mode where you configure your UI.

Why is this not in the same commit as the related unittest, as is common practice?

The description says this depends on D8773 which isn't approved.

Thanks, now it compiles! I'm not sure it's working perfectly though. I updated D25447 to use this, but it's claiming that my indexer is suspended, while balooctl status disagrees.

please send me a list of urls that don't have https: and I'll add them to the whitelist

Fix "+" in keywords to not act as a word break.

@apol we have a duplicate test in breeze-icons now; is this still relevant?