bug#6264: 23; scrolling up erratic when line fractions visible at bottom of window

bug#6264: 23; scrolling up erratic when line fractions visible at bottom of window

[David Reitter]

[-- Attachment #1: Type: text/plain, Size: 1510 bytes --]

Moving down in a buffer does not work well at all when the last line in the window is not fully visible.

The problem occurs in a wide variety of conditions, but here is a test case that works for me:

Emacs -Q  [23 branch build, NS port]
(setq scroll-conservatively 100)
(setq line-spacing 0.6)
(modify-frame-parameters nil '((height . 20)))

Insert some 40 lines of text into the buffer.

Now set point somewhere near the beginning of the buffer, and press <down> repeatedly to scroll the buffer.

You will see at some point that the text is scrolled up by more than one line so the cursor moves vertically up and down. Screenshots down-0 and down-1 illustrate this.

What should happen is that the window scroll so that the baseline of the last text line stays the same, analogous to what happening when we're at the top end of the window, pressing <up>.

NOTE: the above works with the default font (Monaco) on my system; on other systems or with other ports, slightly different line-spacing would apply.


Another, perhaps related problem:
Bottom line (before scrolling starts): here we have a buffer with lines that have different heights (AUCTeX).  
We've got scroll-margin at 0, scroll-conservatively at 10000,  scroll-up-aggressively nil.  The attached screenshot (bottom-half-line) illustrates the problem: the last line is half-visible (that's OK), but the window scrolls too late.

The problem can be seen in the above Emacs -Q session with (setq line-spacing 1.2).
 



[-- Attachment #2: bottom-half-line.png --]
[-- Type: image/png, Size: 16957 bytes --]

[-- Attachment #3: down-1.png --]
[-- Type: image/png, Size: 69149 bytes --]

[-- Attachment #4: down-0.png --]
[-- Type: image/png, Size: 66877 bytes --]

bug#6264: 23; scrolling up erratic when line fractions visible at bottom of window

[Alan Third]

David Reitter <david.reitter@gmail.com> writes:

> Moving down in a buffer does not work well at all when the last line
> in the window is not fully visible.
>
> The problem occurs in a wide variety of conditions, but here is a test
> case that works for me:
>
> Emacs -Q  [23 branch build, NS port]
> (setq scroll-conservatively 100)
> (setq line-spacing 0.6)
> (modify-frame-parameters nil '((height . 20)))
>
> Insert some 40 lines of text into the buffer.
>
> Now set point somewhere near the beginning of the buffer, and press
> <down> repeatedly to scroll the buffer.
>
> You will see at some point that the text is scrolled up by more than
> one line so the cursor moves vertically up and down. Screenshots
> down-0 and down-1 illustrate this.

Hi David, I can't reproduce this in Emacs 25, so hopefully it's been
fixed some time in the last six years.

As always, I'll close in a few days if I don't hear otherwise.
-- 
Alan Third

bug#6264: 23; scrolling up erratic when line fractions visible at bottom of window

[Stefan Kangas]

Alan Third <alan@idiocy.org> writes:

> David Reitter <david.reitter@gmail.com> writes:
>
>> Moving down in a buffer does not work well at all when the last line
>> in the window is not fully visible.
>>
>> The problem occurs in a wide variety of conditions, but here is a test
>> case that works for me:
>>
>> Emacs -Q  [23 branch build, NS port]
>> (setq scroll-conservatively 100)
>> (setq line-spacing 0.6)
>> (modify-frame-parameters nil '((height . 20)))
>>
>> Insert some 40 lines of text into the buffer.
>>
>> Now set point somewhere near the beginning of the buffer, and press
>> <down> repeatedly to scroll the buffer.
>>
>> You will see at some point that the text is scrolled up by more than
>> one line so the cursor moves vertically up and down. Screenshots
>> down-0 and down-1 illustrate this.
>
> Hi David, I can't reproduce this in Emacs 25, so hopefully it's been
> fixed some time in the last six years.
>
> As always, I'll close in a few days if I don't hear otherwise.

It seems like this was never closed.  Closed now.

Thanks,
Stefan Kangas