Tested with jdk1.3.1 ea_build_b08 on Solaris7 Sparc and Solaris8 X86.
JIS X 0212 Kanji characters' width is thiner than others in SansSerif/plain and Dialog/plain with 22 point size or bigger and both of antialias ON and OFF.
This phenomenon is reproducible in Kestrel Fcs, therefore this is not a regression.
1. Launch Font2DTest ($JAVAHOME/jdk1.3.1/demo/jfc/Font2DTest/Font2DTest.jar)
2. Select Dialog or SansSerif.
3. Input 25 in a text field for Size.
4. Select Other from Unicode Range.
5. Input 4e00 in a text field for Unicode Range.
6. Select one of the method from DrawMethod except drawBytes.
Then, you can see the ploblem.
for exapmple, the Kanji character which code is one of 4e02, 4e04, 4e05... are thiner.
This problem is not reproducible with Serif, DialogInput and Monospaced.
reproducible with jdk1.4 beta b44.
4e02, 4e04, 4e05, 4e0c, 4e12, a part of 4e1f, 4e23, 4e24, 4e2b, 4e2e, 4e2f, 4e30, 4e35, 4e40, 4e41, 4e44, 4e47, 4e51, 4e5a, 4e5c, 4e63, 4e68, 4e69, 4e74, 4e75, 4e79, 4e7f, a part of 4e8d, a part of 4e97, a part of 4e9d, 4eaf, 4eb9, 4ec3, 4ed0, 4eda, 4edb, 4ee0, 4ee2, a part of 4ee8, 4eef
are thinner on Solaris, which don't change a lot between being Plain and Bold on Win NT4. See the bug ID 4393264.