SUMMARY When tagging already detected faces (that is, digikam already has the rectangle over the face of several pictures, but without a name yet), sometimes instead of writing the name on the text field, digikam interprets the keystroke as a keyboard shortcut, despite the cursor being a vertical bar blinking in the text field. Specifically, it happens when tagging a face in a second picture after faces in a previous picture have been tagged. I hope it's not too difficult to understand the way I explained. It's actually quite simple once you have seen it in action. STEPS TO REPRODUCE 1. Scan several pictures in a folder for faces. (via right-click, scan for faces). 2. Open the first one (double click), click on the text field under a detected face, and write the name. Click "Confirm". 3. Hover over to another face, or go to the following picture where a face is. 4. Observe how the cursor on the face rectangle text field is blinking. Start typing the name. OBSERVED RESULT Keyboard shortcuts (e.g. going to pictures named with the letter you just typed in the album) are interpreted instead of writing the name of the person. EXPECTED RESULT If the textbox is selected and the cursor is a vertical bar that blinks, it should write the name of the person, instead of using keyboard shortcuts in the album. SOFTWARE/OS VERSIONS digikam-6.2.0-git-20190613T125237-qtwebkit-x86-64.appimage in Ubuntu 18.04 LTS.
Please make a screencast, I can not imagine what you mean. Tested here with a Windows version, nothing to reproduce. From a technical point of view, the input fields have nothing to do with each other, are newly created objects. It would be nice if you would create native digiKam versions on Ubuntu, The AppImage currently has a problem with locale Settings, which leads to several problems. String or date matches are not correct. Please start the AppImage with "LANG=C /path/to/digikam.appimge". Are there any differences? Maik
I made a quick screencast. https://imgur.com/a/GnTDxXe In the video, first I enter a picture with detected faces, and I type the name of the face that already has the focus (Arya). As you can see, I can enter her name just fine. Then I switch to another one in the same picture (Jon Snow). So far so good. However, when I change to another picture, and I start typing the name of the person who has the focus (in this case, "Grey Worm", at the right of the picture), instead of seeing the text, digikam goes to the first picture of that album that starts with the letter I just typed. In this case, "G", so it goes to a file called "Game-of-thrones.jpeg", you can see the filename at the bottom left corner. I hope now it's a bit more clear. If I manually click on the text field I want to input before writing, then it works just fine. But otherwise, it is misleading because a seemingly active text field is not actually "active". And yeah, I had thought about compiling and creating a digikam repository for ubuntu, so users could install the latest versions using apt-get command, but it's not a very straightforward process and I have to look into it. Ideally, it should be as automatic as possible. I'll set up a virtual machine and start experimenting soon.
Sorry, I forgot to mention, changing the locale (with LANG=C) does not change that behavior. I don't think it has anything to do with it. PS: I also tried to add captions to the screencast, but I didn't find an easy way. Any recommendations for linux software that can do that easily?
I can not reproduce the behavior here on openSUSE Linux or under Windows. Are there any global key shortcuts defined on Ubuntu? Do I understand it correctly, if you press "G", it jumps to first image in the list? In the preview, the cursor keys, image up/down, ESC and "T" defined, nothing else. What happens if you open a image in the preview without facial regions and test all letters? Maik
If I press "G" on the Thumbnail or Preview view, a picture in the album whose filename starts by G is selected. Same for other letters. For instance, I can switch between game-of-thrones.jpeg, 2604185h688.jpg, and b3dbddb3-game-of-thrones-season-8-episode-4.jpg by pressing the keys "G", "2" and "B". In the preview view (where the bug seems to happen), it only jumps from picture to picture if the Thumbbar (Control+T) is "selected"/"focused". If you click on the picture preview itself, the thumbbar is no longer "selected", and selecting files by pressing those keys is not enabled anymore (and I presume face names can be entered just fine). So I think that the issue is that, when you switch from one picture to the next in the thumbbar, the thumbbar is still "selected"/"focused" instead of the picture itself, and even if you see a vertical cursor blinking on the face name text field, typing a name will act on the thumbbar (by changing the picture), not on the text field. To solve it, I think either the picture preview gets the focus automatically, or do not show a cursor blinking on the text field as if it is waiting for text to be written. (do you follow what I mean?)
The main problem is that selecting the images in the icon view or preview with the first letter is not implemented or wanted at all. I am at a loss. Maik
That is weird. I'll try it in my windows partition, to see if it also happens, and report back.
I have to take something back, the first letter works here too... Maik
It also works in Dolphin, I have never noticed it... Maik
Git commit 7e212b334cfccfe7be24a26538a11c4afe14d6e6 by Maik Qualmann. Committed on 22/06/2019 at 05:25. Pushed by mqualmann into branch 'master'. set the focus to preview if we have facial regions FIXRD-IN: 6.2.0 M +2 -1 NEWS M +6 -0 core/utilities/facemanagement/facegroup.cpp https://invent.kde.org/kde/digikam/commit/7e212b334cfccfe7be24a26538a11c4afe14d6e6
Well, I see that you already fixed it, but selecting a picture using the first letter also happens in the Windows version.