Created attachment 112757 [details] Document samples. The detailed description with the git clone that caused the problem MIGHT reside at (About 44MiB) https://temporary.softf1.com/2018/bugs/2018_05_12_Umbrello_bug_Report.tar.xz The summary is that the deployment diagrams of the ruined document are not editable. Nor can the diagrams be renamed. The first version of the document, which did NOT have the bug symptoms, got created with the Umbrello version 2.24.3 on a 32bit https://xubuntu.org/ with an "uname -a " of Linux taim 4.15.0-20-generic #21-Ubuntu SMP Tue Apr 24 06:15:38 UTC 2018 i686 athlo n i686 GNU/Linux The Umbrello 2.24.3 originates from the xubuntu.org official package collection, id est is probably some mainstream version. The the first version of the document was fine, no problems with the deployment diagrams. When the first version of the document got loaded into the custom-compiled Umbrello version, 2.28.70, the custom-compiled Umbrello ruined the document and the deployment diagrams of the ruined document became uneditable with neither of the Umbrello versions. The custom-compiled Umbrello version, the 2.28.70, is able to create and edit deployment diagrams that were created with it. The functional, freshly created diagrams can reside in the same document with the non-editable diagrams. The source of the 2.28.70 is a clean clone of the Umbrello official source repository.
May be there is some relation with the https://bugs.kde.org/show_bug.cgi?id=321111 titled "Umbrello doesn't display all diagrams from XMI file."
May be there's a relation: https://bugs.kde.org/show_bug.cgi?id=161183
May be there's a relation: https://bugs.kde.org/show_bug.cgi?id=123159
In your attachment, in the files created_with_Umbrello_version_2.24.3_on_x86.xmi and created_with_Umbrello_version_2.28.70-ef605784f_on_x86.xmi there are only two deployment nodes (computer_A1_c1, computer_A1_c1 resp. computer_B1_1, computer_B1_2) and only one deployment diagram (deployment_diagram_A1 resp. deployment_diagram_B1). In the file deployment_diagrams_creasion_2.24.3_and_ruined_with_v_2_28_70.xmi there are many deployment nodes, with names starting e.g. block_RaspberryPi_like_computer_based_compute_node_..., computer_..., module_... and in the list view there are three diagram proxy objects, FPGA_with_array_of_MCU_modules, FPGA_with_Pi_based_expansion_cards, MCU_clusters but these are only the proxy objects. The actual diagrams to which the proxies refer are not in the file. The problem I have is that the first two files and the "ruined" file are very different in content. Do you have available the file deployment_diagrams_creation_2.24.3 before it was ruined by v_2_28_70 ? How had you created the diagrams FPGA_with_array_of_MCU_modules, FPGA_with_Pi_based_expansion_cards, MCU_clusters ? Via copy/paste or via listview -> right-click on Deployment View -> New -> Deployment Diagram... ?
(In reply to Oliver Kellogg from comment #4) >... > Do you have available the file deployment_diagrams_creation_2.24.3 > before it was ruined by v_2_28_70 ? > > How had you created the diagrams > FPGA_with_array_of_MCU_modules, FPGA_with_Pi_based_expansion_cards, > MCU_clusters > ? > Via copy/paste or via listview -> right-click on Deployment View -> New -> > Deployment Diagram... ? >... I'm sorry, but I really do not remember any more. I do know that I do tend to use the Copy-Paste functionality a lot and I might have used some old file, made edits and then used some "Save as" type of functionality. At least that's my usual usage pattern. All I remember for sure is that to avoid the various problems I have to use my own "fork"(copy with slight edits that I do not remember) to avoid the problems and that the files between that "fork" and mainstream Umbrello are not necessarily compatible. It may sound out of topic here, but as of 2022_11, at 41 years of age, I still do not know, how to create a operating system independent GUI in any technology other than some old form of HTML+CSS+JavaScript+Ajax+http_server, because Java versions are not backwards compatible, FreePascal/Lazarus is not totally backwards compatible and heaven forbid any C++ GUIs, where I have no idea, how to properly handle Unicode. The wchar and alike are just not convincing for me in situations, where even a small text that has only 100 characters, has a mixture of characters from very different regions of the Unicode table. Let's say, a combination of Latin characters, hieroglyphs, smileys and math symbols. An array of bytes per character seems more believable for me from reliability point of view, unless each text has its own, custom, alphabet that is described at some hashtable that matches Unicode to the custom alphabet character code. I really admire the bold people, who take up the enormous task of creating a GUI in C++, be it Qt, WXwidgets, FoxToolkit or any other C++ GUI library. And, yes, it is really pleasant to use a fast C++ based GUI. Basically, I'm piggy-backing on other people's C++ GUI work, including that of the Umbrello developers. Sorry for not having any additional sample data.