Markus> I wanted to point out once more that the "pretty good" algorithm
Markus> means that it is fast but not conformant to the Unicode BiDi
Markus> algorithm. It approximates it, but gets some things wrong, and
Markus> Mark has not implemented explicit codes so far.
Markus is right. The earlier claims to PGBA working like the Unicode
algorithm were due to testing with too few examples. It is not Unicode
comformant.
Markus> Also, the location for the IBM Classes for Unicode (ICU) should be
Markus> given as
Markus> http://oss.software.ibm.com/developerworks/opensource/icu/ rather
Markus> than the alphaworks server that we used to use.
This has been changed on my page.
Markus> Curious only that some of the results from the "Unicode Reference
Markus> Code" do not match the ICU results, which were cross-checked with
Markus> the Java reference implementation that is now part of the standard
Markus> and is going to go onto the CD. We will look into this, or maybe
Markus> someone else (Mark? Asmus?) can look at it, too.
This is curious. I tested the output with both the C and C++ programs (bidi.c
and bidi.cpp) provided at the Unicode site that Asmus provided.
http://www.unicode.org/unicode/reports/tr9/BidiReferenceCpp
I did not test with the biditest.cpp because it won't compile as it is
(missing includes, missing Native_FLAG_DIR* flags, invalid ANSI variable
binding, and missing parameter cast).
I will use the Java reference implementation to check the results of the C and
C++ code.
-----------------------------------------------------------------------------
Mark Leisher
Computing Research Lab I have never made but one prayer to God,
New Mexico State University a very short one:
Box 30001, Dept. 3CRL "Oh Lord, make my enemies ridiculous."
Las Cruces, NM 88003 And God granted it. -- Voltaire, letter
