https://bugs.kde.org/show_bug.cgi?id=375703
Bug ID: 375703 Summary: Moving grouped images into another album removes groups Product: digikam Version: 5.5.0 Platform: Appimage OS: Linux Status: UNCONFIRMED Severity: normal Priority: NOR Component: Usability Assignee: [hidden email] Reporter: [hidden email] Target Milestone: --- When moving a set of grouped images into a different album, all images are moved (so this is not #294578) but the group property is not kept. In the new album I have to group all images again. I would expect the groups to be moved alongside with the images and all other metadata. -- You are receiving this mail because: You are the assignee for the bug. |
https://bugs.kde.org/show_bug.cgi?id=375703
Maik Qualmann <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[hidden email] --- Comment #1 from Maik Qualmann <[hidden email]> --- Yes, resolving the group is currently a workaround for the same or as copied recognized images. Otherwise, the groups would double in the source folder and would be linked. Because the leader of the group may not be copied first, it is difficult to recreate the group during the copying process. Maik -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
Maik Qualmann <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |CONFIRMED Ever confirmed|0 |1 -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #2 from Jens <[hidden email]> --- Is there no clean solution for this? AFAIK the groups only refer to the images anyway (not the paths). so if images are moved, only the image path must be updated. Group information in the database could stay completely unchanged. Right? -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
Mario Frank <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[hidden email] --- Comment #3 from Mario Frank <[hidden email]> --- Hi Jens, I just uploaded a patch in https://bugs.kde.org/show_bug.cgi?id=374591 that, apart from garbage collection also leads to generating less junk. Meaning, if files are moved/renamed, the old images table entries are reused. This way, the image id of the image is preserved. This could solve your problem. This patch does not work for images that were already moved before applying the patch. You can test it if you compile from source and apply the patch. But be advised: The patch is in testing phase. If you do not run the garbage collector in maintenance, nothing bad should happen to your databases. But just to be sure, backup your databases before you test. -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #4 from Jens <[hidden email]> --- Oh, that is perfect. Garbage collection was on my list too - especially since some garbage has accumulated while developing iphoto2xmp: https://github.com/jensb/iphoto2xmp :-) Thank you! Will try as soon as there is an appimage. -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
[hidden email] changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[hidden email] --- Comment #5 from [hidden email] --- New AppImage will be ready tomorrow morning, not before... Gilles Caulier -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #6 from [hidden email] --- New 5.5.0 AppImage is done with garbage database collector patches. Uploading to GDrive is under progress. It will be online in few minutes at usual place : https://drive.google.com/drive/folders/0BzeiVr-byqt5Y0tIRWVWelRJenM New database Garbage Collector options are there : https://www.flickr.com/photos/digikam/32549923912/in/dateposted-public/ https://www.flickr.com/photos/digikam/32549923632/in/dateposted-public/ Gilles Caulier -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
Maik Qualmann <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[hidden email] --- Comment #7 from Maik Qualmann <[hidden email]> --- *** Bug 376014 has been marked as a duplicate of this bug. *** -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #8 from Maik Qualmann <[hidden email]> --- I have for a test the workaround code removed. https://cgit.kde.org/digikam.git/tree/libs/database/item/imagescanner.cpp#n393 But the problem with the wrongly linked grouped image IDs is also not fixed with the garbagecollection branch. Maik -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #9 from Jens <[hidden email]> --- Is this (wrongly linked image IDs) an issue that should concern me? what exactly happens? Thank you :) -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #10 from Mario Frank <[hidden email]> --- (In reply to Maik Qualmann from comment #8) > I have for a test the workaround code removed. > > https://cgit.kde.org/digikam.git/tree/libs/database/item/imagescanner. > cpp#n393 > > But the problem with the wrongly linked grouped image IDs is also not fixed > with the garbagecollection branch. > > Maik Maik, I just compiled the current state of the garbage collection branch and did the following: 1) Define 3 images as grouped 2) move the group to a different directory Result: The images are now located in the target directory and are still grouped. My garbage avoidance for moved items works like follows: If items are moved, I set the album to the target album. Thus, the ImageScanner should not detect the moved image files as new and thus also should not detect the images as identical. Moreover, the images entries stay intact and the image relation grouped should not be touched. I can not reproduce the problem in the current state of the garbage collection branch. Can you both test against this branch? -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #11 from Mario Frank <[hidden email]> --- I have to correct myself. Moving the group back suddenly destroyed the group relation... That's odd. -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
[hidden email] changed: What |Removed |Added ---------------------------------------------------------------------------- Component|Usability |AlbumsView-Group -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #12 from [hidden email] --- Jens, One point : Which database type do you use ? sqlite or mariadb ? Mario, Problem reproducible here too (sqlite) Gilles Caulier -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #13 from Maik Qualmann <[hidden email]> --- Mario, for testing you must remove the workaround code: https://cgit.kde.org/digikam.git/tree/libs/database/item/imagescanner.cpp#n393 Yes, it now works with moved images within digiKam. But not with a copied group or when the same grouped images are added with import. Also externally copied grouped images with the file manager makes problems. Maik Maik -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #14 from PaulK <[hidden email]> --- I am using MySQL on Ubuntu 16.04 -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #15 from Mario Frank <[hidden email]> --- Okay, I thought about the problems for some time. I try to resume the expected behaviour: 1) Moving a group of items should preserve the grouping relation on the moved items (Bug description by Jens) - This works in garbage collection branch without workaround. 2) Moving an item that is grouped should preserve its grouping relation - This also works in garbage collection branch without the workaround. 3) Copying a group within digiKam should also copy the grouping relation for the newly generated items - This is not done. 4) Copying an item of a group within digiKam - This is not clear. Should the new item belong to the group? 5) Importing images into digiKam that are identical to grouped items - Should the imported images be grouped automatically? 6) What do you mean with > Also externally copied grouped images with the file manager makes problems. I assume that you mean that some user copies grouped images that are imported in digiKam with a file manager. But this then also makes problems when grouped images are moved with a file manager. I give here my opinion to 3-6 and I am open for discussion. I will try to make my arguments as precise and founded as possible. I would state the following for 3 and 4: Copy means copy and we should give the user what can be expected by the terms we use. Since we copy all image properties, the grouping relation should be copied in principle, too. But the problem here is: What does the user expect? Should the copies have a grouping relation with the source group leader or should the copied group be a new group with the copy of the old group leader being the new group leader? I cannot answer this question. Only the users can. So the default would be for me to not set group relations for copied items and (potentially) give the user an option to decide about the relations to established. I would state the following for 5: If a user imports some images into digiKam, we cannot know whether he wants to establish a group or add the images to a group. Also, the grouping relation is existent in digiKam but not in file managers - at least not that I know. Thus, the imported images should not be grouped automatically. We could ask the user or give an option as import. But I would not take the decision from the user. I would state the following for 6: I like good user experience. But this case is a clear misuse for me. One cannot expect some complex software to work properly when tampering with the resources managed by the software. If you just copy an image tag relation in our database, you will also have unexpected behaviour if our constraints do prohibit that. And this also holds for other software projects. If you copy some bundles from a OSGI container, he will probably yell at you to stop tampering with his internal resources. Probably with a crash. But we cannot prohibit changes in the file system level. We cannot even detect them if digiKam is not started. This is - at least for me - something that should be stated in the documentation. To conclude: cases 3 to 5 are no bugs for me. I think they are usability issues and we could try to give the user the control about what is done with the group 1) establish a new group with copy of group leader being new group leader 2) link to existent group leader 3) do nothing (default) case 6 is misuse and should be documented. Opinions? All the best Mario -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
Wolfgang Scheffner <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[hidden email] --- Comment #16 from Wolfgang Scheffner <[hidden email]> --- Mario, I agree to 1) and 2): Moving a group or a single items should preserve the group relation. With 3) I don't quite understand the remark at the very end of your comment: > 3) do nothing (default) However, I agree to > 3) Copying a group within digiKam should also copy the grouping relation for the newly generated items but of course it should be a completly new group independent of the original. 4) Here I would vote for not establishing relations to the original group. We could in addition implement an option to keep the copy in the group but I don't like the idea of too many options/settings. 5) I would (as a user) never expect anything to be grouped automatically at import. I why should we group identical images anyway? That's what Find Duplicates is for. And here applies the same as under 4): I don't like the idea ... 6) I agree absolutely. Other than that I am of the opinion that improving metadata handling is the most important issue in digiKam. -- You are receiving this mail because: You are the assignee for the bug. |
In reply to this post by bugzilla_noreply
https://bugs.kde.org/show_bug.cgi?id=375703
--- Comment #17 from Jens <[hidden email]> --- I agree to the previous points. But I have one thought I'd like to share. We are getting to the point where "groups" of images become basically identical to sub-albums (subfolders within an album) except for the fact that they are displayed within the main folder and sorted into the parent's images, not separately. So do we *want* the difference between grouped images and subfolders/albums to exist at all? Is there a difference? Does there need to be a difference? For me, - grouped images represent images that are very similar (e.g. taken with my camera's "machine gun mode" to catch the right moment) but I almost never want to see all of the images. They are part of an event (album) and a chronological series of images. - (sub)albums represent "subevents" of their own, part of a larger event (e.g. one day's tour within a vacation) with different and unique images that all want to be viewed (and some can be grouped). But both could be used for both purposes with very little change in the UI. Both subalbums and groups can have a "master image" (thumbnail) and both subalbums and groups can be opened and closed. Maybe the whole idea of grouped images needs to be rethought? Maybe we can solve many of the grouping issues by just redefining "groups === subalbums"? Very little would need to be done, IMHO: - Include an option for each album (and a global default) to display subalbums in the main thumbnail view (with a folder icon or the defined album thumbnail) as an image within the other images - allow ratings, tags, etc. to be applied to albums (which would effectively apply them to all contained images) Is this maybe worth a second thought? It would make the whole concept a lot simpler and - I think - solve a lot of the consistency issues that the concept of grouping images has right now. -- You are receiving this mail because: You are the assignee for the bug. |
Free forum by Nabble | Edit this page |