Created attachment 121910 [details]
Document with formula and screenshot of bad and good rendering
Open attached document. Notice, that the brackets are far to thick. The document contains an image of the correct rendering.
Build ID: a933e01a54f08132c2d8699f7c6851a8b493d5dc
TinderBox: Win-x86@39, Branch:master, Time: 2015-08-04_06:10:12
Locale: de-DE (de_DE)
Build ID: c614e711136205252ac2c72f9772c718dafc471e
TinderBox: Win-x86@62-merge-TDF, Branch:MASTER, Time: 2015-08-12_20:36:06
Locale: de-DE (de_DE)
While both renderings are bad (scaling brackets to make them bigger is never a good idea, it should be done using glyphs for different sizes like TeX does), but I think the new behaviour is a much more improvement. Thinning the scaled bracket into oblivion is a much worse behaviour than making it thicker.
This seems to have begun at the below commit.
Adding Cc: to Tor Lillqvist; Could you possibly take a look at this one? Thanks
7f0371ad242095657660bb4862bcdfa4a28b4e2c is the first bad commit
Author: Norbert Thiebaud <email@example.com>
Date: Tue Aug 11 23:27:51 2015 -0700
:040000 040000 245203c6558658daaa3c2c56bc92bb8371a13f47 c660e9c56aa230b617b462b941c61f38ef7c0a27 M instdir
author Tor Lillqvist <firstname.lastname@example.org> 2015-08-12 05:18:50 (GMT)
committer Tor Lillqvist <email@example.com> 2015-08-12 06:14:20 (GMT)
commit 4667db065d34193d99bce82f7e8f3b20a03ecade (patch)
Created attachment 121962 [details]
screen capture from LO 220.127.116.11.0+
Not reproducible for me with LO 18.104.22.168.0+ built at home under Ubuntu 15.10 x86-64.
I do not use OpenGL nor GTK3 in LO.
Best regards. JBF
I would not judge the image Regina inserted as "Good rendering" in her example attached to the description satisfactory. (May this be due tu unintentional clipping?)
I meanwhile tested with "Version: 22.214.171.124 (x64)
Build ID: ecd3574d51754b043f865cf5bafee286d24db7cc
CPU Threads: 4; OS Version: Windows 6.29; UI Render: default;
Locale: de-DE (de_DE)"
In addition: UI en-UK
No improvemenmt: Ugly brackets and braces unchanged as compared with Regina's example.
Tor reverted the incomplete Uniscribe implementation on Windows with
For now, that has cleared the regression here. But Frédéric's and Khaled's comments on bug 32362 on need to rework Math's formula canvas to construct the brackets, parenthesis and boxes with glyphs for multiple codepoints (assume we'd work with codepoints 0x239B -> 0x23B9 from the Unicode Miscellaneous Technical block).
Reverted to *less* objectionable stretching with reversion of UniscribeLayout support and reintroduction of SimpleWinLayout
first landed in master
Build ID: c81eddbb20c84280aa64c712e34c829380b24527
CPU Threads: 8; OS Version: Windows 6.19; UI Render: default;
TinderBox: Win-x86@39, Branch:master, Time: 2016-01-22_04:19:03
Locale: en-US (en_US)