Manual invocation of trust search

[egbicker]

Addresses slowness in search due to compounding search events by changing the search bar behavior to only begin filtering when Enter key is pressed instead of upon any key press.

The root cause of the slowness was multiple compounding search events generated by the 150 millisecond keystroke delay. That is now eliminated at the cost of having to press enter to initiate search.

Closes #684

[egbicker]

The activate is a good call. The X icon was causing me to have some unresponsive behavior prior to this branch too. If I let it sit and don't close the application, it will eventually clear the search bar.

[jw3]

If there is something strange with the button and it can be removed/hidden, that would be acceptable. A button to clear a single field is of minimal value

[egbicker]

The only way I could find to remove the "clear" icon didn't seem to work before the user began inputting text, so now its getting drawn and erased on each keystroke which I don't like. Not sure what the best option here is

[dorschs57]

The only way I could find to remove the "clear" icon didn't seem to work before the user began inputting text, so now its getting drawn and erased on each keystroke which I don't like. Not sure what the best option here is

Have you tried the icon_release signal instead of the icon press? Its also possible things are locking up because the built in functionality of the clear button is also kicking in at the same time. You can try stopping event propagation by returning a True from you handler I think. You might have to handle both the press and release signals to stop propagation in both places.

If all that fails you can also try setting your own search icon in the secondary position that might override the default behavior so you can handle it the way you want. According to the docs on Gtk.SearchEntry

Note that the search/clear icon is shown using a secondary icon, and thus does not work if you are using the secondary icon position for some other purpose.

[egbicker]

@dorschs57 suggestion to change SearchEntry to Entry removed the clear button. Now the user must backspace and hit enter again to revert to the full trust database display. Working on the test to simulate an enter key programmatically now.

[dorschs57]

@egbicker @jw3 I'm seeing some better performance from the filtering with my changes for async trust loading. Its limited testing right now, but it might be work holding off for my work on #711.

[jw3]

@egbicker conflicts after #750

[egbicker]

@jw3 I'm seeing an issue with the new "check_ancillary_trust" with (left != right)

[jw3]

@egbicker Did something happen when you resolved conflicts? I see only one line changed in this PR.

Assuming this branch is broken, I took from the previous commit and rebased on master to test. Saw seemingly sane behavior.

List the steps to repro the error and Ill test again.

[egbicker]

@jw3 I'm seeing this in my master branch, looks like one of the checks in the rust backend doesn't like what I have going. I was behind on syncing my fork so I just got that update.