Re: Fwd: Malayalam & SF_PROCESS_ZWJ

Hi all,

On 6/19/07, Eric Mader <emader icu-project org> wrote:
Just to be clear, the SF_PROCESS_ZWJ says that the ZWJ/ZWNJ glyphs
should be passed to GSUB processing. For most Indian scripts, it's
sufficient to see the ZWJ/ZWNJ characters during character processing
where they are used to assign the correct features to the surrounding
characters. I think it should be possible to do everything needed for
the Malayalam script without passing the ZWJ/ZWNJ characters to GSUB.
(Though I haven't looked into the particular bug being discussed ;-)

I don't know how to read Malayalam, but I don't see anything in the
section on Malayalam in the Unicode documents (5.0) which would make
SF_PROCESS_ZWJ a requirement for Malayalam rendering.

Talk of developing our own standards makes me very nervous. I think it's
important that we implement the same standard as Windows.
Interoperability must be our most important goal. Otherwise chaos will
result. The thing I hear most often from clients submitting bugs against
the ICU LayoutEngine is "but it works on Windows..."

Users expect things to work the same way on every platform, and they should.

I agree entirely. What has been a dire necessity for a long time is a
set of well defined test cases for all the Indic scripts that contains
all the known corner cases and special cases (chillu in Malayalam,
ra-japhala in Bengali, etc) and which can be tested against various
rendering systems (Pango/QT/ICU/Uniscribe) to ensure that things show
up in the same manner no matter what the underlying mechanism and/or
platform is.

To begin with the efforts, we (the Bengali team) have been using a
very rudimentary test case of our own
for testing the fonts that we have developed. It is based on a list of
possible conjuncts (taken from a spelling dictionary commonly used for
our language) and also based on the experience we have gathered while
working on Bengali and FLOSS for the past five years. If such a list
can be compiled for the other scripts as well, we may have the
beginning of a Indic test suite that has been discussed in various
forums/workshops/conferences/mailing lists for the past four years (at

Warm regards (and sorry for the semi off topic rant)

Sayamindu Dasgupta

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]