Download it now!
Bug 134761 - FILEOPEN ODT: Distribution content over pages completely different (since compared to older versions) since 6.4
Summary: FILEOPEN ODT: Distribution content over pages completely different (since com...
Status: NEW
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Writer (show other bugs)
Version:
(earliest affected)
6.2.0.3 release
Hardware: All Windows (All)
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords: bibisected, bisected, regression
Depends on:
Blocks:
 
Reported: 2020-07-12 20:04 UTC by Telesto
Modified: 2020-07-13 17:43 UTC (History)
1 user (show)

See Also:
Crash report or crash signature:


Attachments
Example file (13.57 MB, application/vnd.oasis.opendocument.text)
2020-07-12 20:05 UTC, Telesto
Details
Bibisect log (3.21 KB, text/plain)
2020-07-12 20:17 UTC, Telesto
Details
Export 62 (2.54 MB, application/pdf)
2020-07-13 13:12 UTC, Telesto
Details
Export 71 (2.33 MB, application/pdf)
2020-07-13 13:12 UTC, Telesto
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Telesto 2020-07-12 20:04:57 UTC
Description:
Distribution content over pages completely different (since compared to older versions) since 6.4

Steps to Reproduce:
1. Open the attached file 
2. Scroll to page 8
3. On top of the page should be 2 allow bullets (at least that the case in 6.0-6.2). Not a whole list

Actual Results:
10-15 arrow bullets the start of page 8

Expected Results:
2 arrow bullets


Reproducible: Always


User Profile Reset: No



Additional Info:
Found in 7.1

and in
6.4

but not in
6.3
Comment 1 Telesto 2020-07-12 20:05:54 UTC
Created attachment 162939 [details]
Example file
Comment 2 Telesto 2020-07-12 20:17:03 UTC
Created attachment 162940 [details]
Bibisect log

Bisected to

author	Jan-Marek Glogowski <glogow@fbihome.de>	2019-07-04 01:45:11 +0200
committer	Jan-Marek Glogowski <glogow@fbihome.de>	2019-07-05 19:38:01 +0200
commit	7dd44125a9c184c21374a47005e303579179805f (patch)
tree	9b55495a8708231f65b98bb0eabda6c86c6f5d4c
parent	4154d281fbecaadf6cd118c00cc6cff929e339a4 (diff)
WIN don't register LO's shared fonts twice
This separates the system registration (AddFontResourceEx) of
LO's shared fonts from the later PhysicalFontCollection
population using EnumFontFamiliesEx.

Calling AddTempDevFont from GetDevFontList creates a new
WinFontFaces, but the later EnumFontFamiliesEx also creates
WinFontFaces for all registered fonts, resulting in duplicate
WinFontFace entries in the PhysicalFontCollection.

Also currently the provided WinFontFace from AddTempDevFont is
much less accurate, compated to the EnumFontFamiliesEx one.

https://cgit.freedesktop.org/libreoffice/core/commit/?id=7dd44125a9c184c21374a47005e303579179805f

-> An additional confirming bibisect would be nice
Comment 3 Telesto 2020-07-12 20:18:06 UTC
If you have some time, would you mind to check the bibisect..
Note: Win only
Comment 4 Telesto 2020-07-12 20:36:07 UTC
Has probably something to do with Font Substitution
Comment 5 Buovjaga 2020-07-13 12:34:49 UTC
(In reply to Telesto from comment #3)
> If you have some time, would you mind to check the bibisect..
> Note: Win only

I don't understand the report. Please include a screenshot of good vs. bad result.

Bibisect result at least seems wrong, but maybe there is something that gives us a different result - maybe I am looking at the wrong page. Screenshots should help with that.
Comment 6 Telesto 2020-07-13 13:12:31 UTC
Created attachment 162972 [details]
Export 62
Comment 7 Telesto 2020-07-13 13:12:51 UTC
Created attachment 162973 [details]
Export 71
Comment 8 Telesto 2020-07-13 13:13:54 UTC
Clearly something with fonts.. first thought.. font substitution change.. but bibisect results in comment 2
Comment 9 Buovjaga 2020-07-13 13:54:16 UTC
Looking at the PDFs, I still don't understand what you mean. Please quote some lines and refer to the pages of the different PDFs.
Comment 10 Telesto 2020-07-13 14:24:43 UTC
(In reply to Buovjaga from comment #9)
> Looking at the PDFs, I still don't understand what you mean. Please quote
> some lines and refer to the pages of the different PDFs.

Side by side view helps

p9 export 6.2 is p8 7.1. And look at distribution of text/images & the font(size)etc..
Comment 11 Buovjaga 2020-07-13 15:05:43 UTC
(In reply to Telesto from comment #10)
> (In reply to Buovjaga from comment #9)
> > Looking at the PDFs, I still don't understand what you mean. Please quote
> > some lines and refer to the pages of the different PDFs.
> 
> Side by side view helps
> 
> p9 export 6.2 is p8 7.1. And look at distribution of text/images & the
> font(size)etc..

From your description:
"On top of the page should be 2 allow bullets (at least that the case in 6.0-6.2). Not a whole list"

I understood that it was about *the same content*, but incorrectly having many more bullets.
Comment 12 Buovjaga 2020-07-13 16:42:43 UTC
I am bibisecting this with win32-6.2
Comment 13 Buovjaga 2020-07-13 17:25:41 UTC
win32-6.2 gives me

https://git.libreoffice.org/core/commit/b96bbb1ff391036cbba832bc22100792880f0fe6
upgrade the internal harfbuzz to 1.8.4

Sounds plausible. Hard to say what is the exact issue that needs fixing here. I certainly have no idea what would be the preferred look. They both look messy to me.
Comment 14 Telesto 2020-07-13 17:43:57 UTC
(In reply to Buovjaga from comment #13)
> win32-6.2 gives me
> 
> https://git.libreoffice.org/core/commit/
> b96bbb1ff391036cbba832bc22100792880f0fe6
> upgrade the internal harfbuzz to 1.8.4
> 
> Sounds plausible. Hard to say what is the exact issue that needs fixing
> here. I certainly have no idea what would be the preferred look. They both
> look messy to me.

Hmm, lets focus on the source file, seems a better approach: attachment 134746 [details]

And this gets really interesting :-). The has with 6.2.9 an image called Görüntü11. The same image isn't present in 7.1.0.0.alpha0+.  

One of many weird 'tracking changes' issues. For full list of bugs I created based on this file, see the see also of bug 134746 (source doc)

And no clue what you bibisected: I see no difference between 6.2 and 6.3.. 
So where surely on a different path :-)