Created attachment 198079 [details] An ODT that describes + displays the issue (image + paragraph) Selecting the whole of an image caption (eg for rename) also selects the image. If the caption is then deleted the image itself will also be deleted. Further, if a paragraph-with-inline-border is below the image, and that paragraph (alone, not including the image) is selected in it's entirety, the image is also selected. The odt + pdf that illustrate the problem are also available from Python3docs (bug2.odt + bug2.pdf at https://github.com/alexkemp9/Python3docs).
https://help.libreoffice.org/latest/en-US/text/swriter/guide/captions.html?DbPAR=WRITER#bm_id3147691
Confirm with Version: 25.2.0.0.beta1+ (X86_64) / LibreOffice Community Build ID: a49bea44b7989afd850e9a4a56c2c8a0d1a1ff37 CPU threads: 4; OS: Linux 6.8; UI render: default; VCL: gtk3 Locale: cs-CZ (cs_CZ.UTF-8); UI: en-US Calc: threaded Works in Version: 6.1.0.0.alpha0+
This seems to have begun at the below commit in bibisect repository/OS linux-64-6.5$. Adding Cc: to Michael Stahl ; Could you possibly take a look at this one? Thanks 1771c50f6b4ee8a1737edc85a2ff7f27bcff28a4 is the first bad commit commit 1771c50f6b4ee8a1737edc85a2ff7f27bcff28a4 Author: Jenkins Build User <tdf@pollux.tdf> Date: Fri Dec 6 14:04:55 2019 +0100 source 91b2325808a75174f284c48c8b8afc118fad74e4 84576: tdf#121300 sw: consistent fly at-pargraph selection | https://gerrit.libreoffice.org/c/core/+/84576
1. there is only one paragraph in the text frame that contains the image, so if you select the whole paragraph then the entire text content of the text frame is selected, so all objects that are anchored in the text frame are selected. 2. the text frame is anchored to the paragraph "print(x, y, z) # prints blue green blue". if you select from the first letter to the last letter of the paragraph, then it is not selected, but if you then extend the selection outside of the paragraph, then the "whole paragraph" is selected and every object anchored in the paragraph is selected too. this is working as designed.
Michael Stahl: "this is working as designed." Seriously? It is not possible to delete the caption for an image without also deleting the image, and that is by design? Good grief.
Created attachment 198272 [details] Extended descriptions, demonstrations + comments as ODT + PDF Both ODT + PDF are available at GitHub Python3docs: https://github.com/alexkemp9/Python3docs
It seems that the problem is in the image anchored as a paragraph. By changing it to character, it allows you to easily select to delete all the text in the paragraph, not including the image.
This comment concerns the secondary part of the original bugreport, which is of interference by the image+caption frame on text paragraphs below, and therefore independent to, the image. 1) See also bug2+.odt + bug2+.pdf + bug02+24.8-content.xml at https://github.com/alexkemp9/Python3docs. 2) This issue is #4 within bug2+.odt 3) It has been tested by me under 24.8.3.2, 7.4.7.2, 7.2.3.2, 6.1.0.3, 5.0.6.3, v4.0.0.3 + v3.3.0.4; all are subject to this bug. 4) The xml within content.xml for this specific page (#10) is malformed. 5) In addition, the 1st paragraph of the independent text has been injected *inside* of the image+caption frame. HTH
There wasn't any text in the paragraph that the image is anchored to. Originally it was an empty paragraph. The so-called "text in the paragraph" is the top-text of 3 paragraphs of text introduced *below* the image+caption. The outer frame of that image+caption then expropriated the top text paragraph & incorporated the bare text between the end of the outer-frame & the end of the original paragraph, whilst at the same time throwing away the surrounding paragraph xml. The 2 remaining paragraphs are outside, below the the end of the original paragraph. The whole thing looks like this (as a schematic): Example page#4:- image+caption + 3 lines of Code Box 2 text =========================================================== ⎡---<paragraph#1> | ⎡---<frame#1(Picture+Caption) | | ⎡---<text-box> | | | ⎡---<paragraph#2> | | | | ⎡---<frame#2(Picture) | | | | | ---<image /> | | | | ⎣---</frame#2> | | | | ⎡---<caption>(7 xml lines) | | | | ⎣---</caption> | | | ⎣---</para#2> | | ⎣---</text-box> | ⎣---</frame#1> | x = "blue" ⎣---</para#1> ---<paragraph#5>y = "green"</para#5> ---<paragraph#6>z = x</para#6> I call that an abortion, but I understand that the official view is that "this is working as designed"™.
I'm going to move discussion of "problem#4 within bug02+" into bug03, since those 2 are identical. It will also continue in a few days time, since I have other work that must be completed at this time.