Unicode rendering problems

[Didi Kanjahn]

> I'm not top posting

Khmer Unicode does not get rendered correctly #1.

A major bug in the rendering of Khmer Unicode was fixed in version 2.13.3. 
With 2.13.3 the first glyph-sequence of each line (also line 6) was now rendered
as can be seen in line 1, 3 and 4 of 'test Khmer Unicode 2-13-3.pdf'. Before
version 2.13.3 all Khmer fonts rendered those lines as in line 2 and 6.

However, in version 2.13.19 there is a new problem with rendering Khmer Unicode
(it started at an earlier version, but I don't know when exactly), leading to
many dotted circles. From that ensuing result, it can be assumed that the
instructions of the Unicode rendering engine (not part of the font) are being
carried out (in particular the shifting of certain glyphs/symbols within a
cluster), but the instructions of the OpenType tables that are part of the font
are not being carried out. Please see 'test Khmer Unicode 2-13-19.pdf'.

Please disregard the last dotted circle in each line for this particular bug
report. See below under 'Khmer Unicode does not get rendered correctly #4' for
details.

All fonts and other files can be made available.
All of the above was done under Windows XP.

The font 'Khmer OS' can be downloaded at:
http://sourceforge.net/projects/khmer/files/Fonts%20-%20KhmerOS/KhmerOS%20Fonts%205.0-%20LGPL%20Licence/KhmerOS_.ttf/download

Unfortunately it was not possible to attach any files, the other fonts and
sample pdf files.

The lilypond code to illustrate the problem can be found in my post on 'Font
selection problem'. I don't know yet whether the Unicode characters came across
intact.



Khmer Unicode does not get rendered correctly #2.

To my surprise, even version 2.13.3, which does render Khmer Unicode correctly
with many fonts, there is at least one font in which even the first
glyph-sequence is not rendered correctly. Please see lines 2 and 6, where one
diacritic below the base line is not placed correctly. This is highly
surprising, because all the fonts used in lines 2, 3 and 4 in 'test Khmer
Unicode 2-13-3.pdf' are _very_ similar in their OpenType instructions (all 3
fonts were created by me).



Khmer Unicode does not get rendered correctly #3.

There is one more rendering problem even with Lilypond version 2.13.3 which
otherwise renders Khmer Unicode well (but see bug report 'Khmer Unicode does not
get rendered correctly #2+4').

In the second but last glyph sequence of line 1 (repeated in line 5), a glyph
that is supposed to be under the glyph group is actually above the base line. It
is not clear why this should affect only this font, but it could be that an
OpenType instruction to move this glyph from its normal position directly below
the baseline downwards is actually inversed to move it the same amount upwards. 



Khmer Unicode does not get rendered correctly #4.

In vresion 2.13.3 and 2.13.19, the last glyph sequence in every line (repeated
in line 7) of 'test Khmer Unicode 2-13-3.pdf' is incorrectly rendered as 'not
permitted' sequence with a dotted circle inserted. This is most likely a problem
with the underlying Unicode rendering system (is it called Pango?) and not with
Lilypond. However, the sequence is legitimate and is an existing word in the
Khmer language.

The file 'test (MS Word).pdf' shows the correct rendering of the lines, created
in MS Word on the same computer on which 'test Khmer Unicode 2-13-3.pdf' was
produced.