More details:

Thanks! I had no idea what was causing the error.

There is a side effect here that AF_SimpleCalculate now returns a localized number which will have group Seperators according to the system locale. See my inline comment about this.

Force consistent locale in calculate text test to avoid
different results.

Update the calculatetexttest and simpleCalculate PDF with
formatting removed from the fields in simple calculate.

Just noticed that this breaks the calculatetexttest as I've accidentally also added Formatting (which Acrobat did automatically) on the calculated fields in that test document.

The test is D13588

Apologies for the delay. Some travel, some sick days and some other work and the weeks fly by ;-)
I'm getting back to this now.

Yes, I plan to add a test for this.

• Simplyfy page lookup in processFormatAction
• Clarify comment about the need to refresh calculated

fields in processFormatAction.

Remove spurious comment

Removed to currency and extended numberToString to handle this, too.

In D13170#270082, @aheinecke wrote:

In D13170#269657, @aacid wrote:

QTest::qWait( 100 ); is bad as you probably already guessed since it's "machine dependant".

Yeah, looked fishy ;-) Although stuff like that is used in part test on other places so i've copied it.

Use QTRY instead of QTest::wait. This is acutally much faster.

In D13170#269657, @aacid wrote:

QTest::qWait( 100 ); is bad as you probably already guessed since it's "machine dependant".

In D13171#269646, @aacid wrote:

I'm not very happy of the internalText method/storage in form, ideally, the text in core/form should always be the internal text, and if there's a formatted text that needs to be different that should only be as part of the widget itself. i.e. the FormLineEdit would do something like

setText( formatText( form->text() ) );

Do you think it could be implemented this way?

Don't eat the MouseRelease event if it is outside the widget

• Fixed precedence of activation action over mouse release action.
• Only activate mouse released if release is inside the widget.
• Fixed check for Poppler 0.65

Include duplicate cals from editFormList and editFormCombo

I'll upload a new patch. (sorry for delaying my okular work, I'm currently still a bit swamped by the efail hype)

Works in Ubuntu 18.04 with default breeze theme:

Ugh, Sorry for overlooking these.

To clarify:

I did not know when adding f0a80a67 that pushing to the undoStack automatically executes the redo function.

• Removed a spurious emit when calling signalAction.
• Updated to API of update poppler patch.
• Fixed leaking the additional actions.

Changed mouseRelease event to fallback to activationAction
Some formatting.

MouseReleased indeed needs some different handling. It only works in the tests for the Button's as they trigger the "MouseReleased" action as the activation action.

Turned out that this actually does not need much code (with some preprocessor macros).

As both Bugs mentioned here have multiple issues (Bug 307304 needs the display java script object, Bug 306818 needs visible read only fields) and I want to avoid muddying the waters, I've built a test PDF that can be used as a basis for a unit test:

Thanks :-)
I'll keep the issues you mention in mind, but my current work is for a customer so I'll have to prioritize issues for them.
But I will also look at some other bugs / issues, like the focus in / out thing and stick around as a KDE developer even after the current work is done. Although my main focus is and will be Kleopatra. ;-)

For the record Key Source (origin) is part of Key / UserID in GPGME 1.11.1

All done :-)

Yes :-) I think so.

Reset-Form Actions are specified in Adobe's PDF Reference as: