In Cricket, Hebrew punctuation marks are displayed in unbalanced positions
and sometimes coincide with the letters they modify. The dot in the shuruk is
on the wrong side of the vav.

There are so many individual problems with the punctuation display that I
have written this report in general terms.

It appears that the current placement of Hebrew punctuation marks is
implemented by an algorithm that is not font-specific and does not take into
account special case letters. The marks appear to be placed according to the
following heuristic:

The borders of a glyph are determined using the font metric, without respect
to the visual center or other visual characteristics of the glyph. Then,
a) accent dots are placed at the vertical and horizontal center of the glyph
b) marks under the baseline are horizontally centered

This heuristic causes the puncutation marks to appear unbalanced or to
coincide with the letters they modify.

In addition, the punctuation marks are displayed coincident with readers
marks (taame hamikra) if both are used together.

The solution to this problem requires the use of a shaping engine similar to
the shaping engine used to display Arabic text. In addition, the fonts must
provide glyphs for the accented versions of the letters that can take accents.

The following explains the requirements for correct display of punctuation
marks in Hebrew text:

1. marks must not coincide with or touch the accented letters
2. marks must be positioned according to graphic expectations

If the first requirement is not met and marks coincide with letters or with
other marks, the text might not be readable as intended. If either requirement
is not met, the text looks sloppy and unpresentable.

There is no way to meet these placement requirements without respect to the
font and without respect to special case letters that are the same for
all fonts. That is, there is no algorithm for correct placement of punctuation
marks that is correct for all fonts.

There are several cases that must be considered.

1. Accented letters (dagesh)

In each font, the position of the accent dot (dagesh) in an accented peh (peh
dagushah) depends on the degree to which the peh is closed. If it is highly
closed, then the dot must centerd inside the closed area. If the peh is
relatively open, most fonts place the dot in line with the line that would
close the peh if the line were extended. The letter tet has a shape reminicent
of peh but rotated clockwise by 90 degrees. Obviously the accent dot in the
tet cannot always have the same vertical position relative to the baseline of
the font as the dot in the peh. A font with a highly close peh will place the
dot in the accented peh higher than the midline and the dot in the accented
tet below the midline. The accent in the vav however, rarely varies from the
midline for any font. However, it must be placed to the left of the vav, which
might or might not require extending the space of the vav.

2. Special case letters

For all fonts, the horizontal position of a holam-haser dot after the letter
lamed must take into account the vertical rise of the lamed on its left side
and not coincide with the lamed. The holam-haser after a lamed is therefore
usually placed over the right corner of the letter following the lamed. For
all other letters the holam-haser can be placed over the upper left corner of
the letter.

The shva after an unaccented chaf in word-final position is usually placed
above the baseline.

Below-baseline punctuation marks for an eyin that extends below the baseline
are displayed to the right of the extension.

3. Punctuation below the baseline

The rest of the punctuation marks are usually displayed below the baseline.
These marks must be horizontally centered under the visual center of letters
they modify, except when readers marks (taame hamikra) are also present, in
which case the punctuation and the readers marks must be horizontally
concatenated and the resulting string must be horizontally centered under the
visual center of the modified letter.