I just got access to a brand new M1 MacBook Air and was able to build and
run a M1 native version of PyLucene. Needless to say, it's fast !
By 'native', I mean not emulated via Rosetta.
For the curious, here are the steps:
Build a native Python 3.9.5 from sources (a python.org binary install
may work as well but I didn't try that)
Install a native arm64 early access JDK 17 build from http://jdk.java.net/17/
That early access JDK is a tar archive that is a self-contained JDK tree
with all pieces required. Unpack it <someplace> and
- export JAVA_HOME=<someplace>/jdk-17.jdk/Contents/Home (ant needs it)
- export PATH=$JAVA_HOME/bin:$PATH
Edit JCC's jcc/helper3/darwin.py and add
JAVAHOME = '<someplace>/jdk-17.jdk/Contents/Home'
at line 29, in the Exception block since /usr/libexec/java_home is going to
fail.
edit JCC's setup.py to remove the /jre/ part of the paths under the
'darwin/home' key in LFLAGS.
Build and install JCC as usual.
Edit PyLucene's Makefile and change the -arch parameter in JCC to -arch arm64
in your Mac OS X configuration.
Build and install PyLucene as usual.
Andi..
