Yes, that seems to work (and simplifies the code a bit).

Updated in accordance with review comments, variable names corresponding to what they actually do.

If you mean to incorporate the tests directly into actionRequestedByUser:

I'm not convinced that trying to show a reading time is that useful. Estimating this will need to take so much into account - it seems easy to divide the length of the message by the user's personal reading speed, but getting an accurate estimate will also depend on the amount of original (non-quoted and non-signature) text, the variation in reading speed between individual messages (junk can be dismissed with just a glance, a party invite may just be skimmed for a date/time, while a technical discussion will have to be read in detail, etc) that an estimate algorithm would need to be trained on a great number of messages before getting anywhere near a realistic result. If the time shown bears no relation to reality then the user will learn to ignore it and there is no point in showing it.

Good idea to have the button, but maybe it should be placed in the header area instead of the body of the message for security reasons (so that it cannot be spoofed by a malicious HTML message).

Ignore the desktop ID and simply use the loop index, adjusted so that the desktop numbers start at 1. This is the same formatting as used by KWin (kwin/useractions.cpp) and the Task Manager applet (plasma-desktop/applets/taskmanager/package/contents/ui/ContextMenu.qml). This should work as intended on both X11 and Wayland.

Thanks for the fix - a much better solution to the HiDPI support.

Thanks for the fix!

@GB_2 thanks for testing; a desktop UUID is obviously not friendly to show to the user, so there needs to be some sort of X11/Wayland runtime check here.

In D22740#512556, @ngraham wrote:

So does this fully fix 410190, or only partially?

Good idea to reduce required dependencies, many thanks.

Confirmed that man:clang(1) now correctly displays the man page with no spurious numbers shown. Would be happy to abandon this review request.

Good point regarding the possibility of decoding other formats.

@mlaurent, are you also happy to accept this revision?

New function setUseProxyInternal() to set proxy state and, if the connection is active, disconnect and reconnect it with the new proxy setting.

Diff updated in accordance with review comments

Tab converted to spaces

KIMAP required version updated

Version number updated

Laurent, do you mean the PIM_VERSION in kimap/CMakeLists.txt (which then sets KIMAP_LIB_VERSION), currently "5.11.40"? I was under the impression that this was managed by the release scripts and all PIM components had to have the same version.
I assume that you don't mean the SOVERSION of the kimap library (which doesn't have a minor version).

See https://bugs.kde.org/show_bug.cgi?id=407685 for IMAP.

Appears to have been already implemented by https://phabricator.kde.org/R244:b2b873d612a018d33866ddfd497f87bb7dd79573

The tool is up to date and working as far as I am aware, but any bug reports or suggestions for additions would of course be welcome.

Use the standard export header

Ping - anyone able to review please?

No need to declare destructor as virtual

Will let the Akregator change soak for a bit, to make sure there are no problems, and then submit it.

Setup and destruction of the QWebEngineProfile moved to WebEnginePage, so MailWebEngineView need know nothing about the profile and MailWebEnginePage only needs to access it via profile().

@dvratil: in light of @anthonyfieroni's comments, should this fix go into WebEnginePage (messagelib/webengineviewer/src/webenginepage.cpp) instead?

There's always a better way to do it...

Yes, doing that seems to work. It's not as elegant an implementation as it could be, because the QWebEngineProfile still needs to be created by MailWebEngineView and passed to the MailWebEnginePage which adopts it as a child. It's not possible to manage the profile entirely within MailWebEnginePage because a profile can only be set when a QWebEnginePage is created, and trying to do that in the constructor:

Sounds a good idea, will investigate.