How to reproduce:
1. Create a new (empty) Writer document.
2. Change the zoom factor to such a value that the page width does not fit to the screen. E.g. 400% (for me).
3. With the horizontal scroller (at the bottom of the screen), scroll to the right so that you see the right edge of the page (the left edge disappears).
4. In the horizonal ruler (in the top area of the window), change the page margin, e.g. from 2 cm to 1 cm.
The margin gets changed BUT the view jumps to the left so that I see the blinking cursor (and don't see whether the margin really was changed and have again to scroll to the right to see it).
I'd expect the action (changing the right page margin) to not to change the view, i.e. I still should see the right edge of the page.
Don't see much of an issue, it is self documenting what has happened. You've made a change to page layout setting a margin, and the edit focus has shifted the vcl canvas view to follow the edit cursor of an empty paragraph.
What happens when you fill the page with text, position the cursor near the end of a line and then apply the margin change?
With recent master nightly, the view pane follows the edit cursor focus. That is expected behavior--follow the edit cursor and pan when the page does not fit in the current view pane. Otherwise you are hunting for the edit cursor.
IMHO => NAB and definitely a => WF for any change.
For me, this a very unexpected behaviour. I do something in one area and a certain context and get thrown to a completely different context. This is not how I understand the principle of the least astonishment.
> What happens when you fill the page with text, position the cursor near the end
> of a line and then apply the margin change?
Depending on the resulting text, the h-scroll position changes so that the insertion mark is visible. This confirms your words "the edit focus has shifted the vcl canvas view to follow the edit cursor of an empty paragraph". But this is not what I'd expect. This changes the context too much. What if I missed the desired margin and want to correct it again? All of a sudden, I'm somewhere else and need some seconds to realize what just happened.
My expectation would be that the visible part of the page does not change after working with the margin ruler -- even if the insertion mark gets out of view. After I move the insertion mark (by typing a letter or by moving it with the arrow keys), I'd expect the insetion mark to become visible because now the insertion mark is my mental focus.
Same issue when you scroll down and resize the bottom margin - the view jumps back to the focus. Use case might be to place shapes at a certain position.
If we change this it has to work in all situations. What should happen when you click at the canvas? The view changes to the cursor now. What if you modify some attribute, let's say the font size?
> Same issue when you scroll down and resize the bottom margin
Yes! Also not what I'd expect.
> What should happen when you click at the canvas?
What do you mean by "click at the canvas"? I'm not familiar with the internal LO terminology (which the expression seems to be).
If you by that mean the page area where one writes the text then it depends whether the click location contains the text. If it does then the insertion mark should be placed where the user clicked (or the nearest possible place). If the page is blank there (e.g. if it's scrolled down and the text ends somewhere above) then I'd probably expect the view to jump so that the insertion mark is visible. The reason is that if I click on the page then my mental focus is on the page (writing text). Hence making the mark visible (and jumping to it if necessary) is justified.
(In reply to fml2 from comment #4)
Then why do you perceive the behavior as incorrect when having zoomed in the view of the page to such an extent that making a margin change reorients the page and then perfectly consistently moves the vcl view port of the canvas to the only meaningful point--the edit cursor?
It would make little sense to hold the vcl canvas view on a part of the page that no longer has any importance to inputting textual content.
> Then why do you perceive the behavior as incorrect...
Because now the focus of my actions is changing the margin, and I wouldn't expect to be taken away from that without me doing something that gives a hint that I want to change the focus.
Such "doing something giving a hint" could be e.g. a click into the writing area. This would be a clear sign that I'm done with changing the margin, and would thus justify making the insertion mark visible (and changing the view).
Obviously we have different opinions here. Let's keep the ticket open and ask for dev advice. It's clearly a minor annoyance but affecting many if not all users.