Persist LLVM IR to .pyc files

[GoogleCodeExporter]

Currently, the compiler emits both CPython bytecode and LLVM IR and
attaches both to all code objects it creates. However, the marshal format
only understands bytecode and drops the LLVM IR on the floor when it saves
.pyc files. Disabling .pyc files entirely seems to slow down regrtest to
unacceptable levels so r323 just checks that the LLVM IR is present when it
tries to run through the LLVM JIT and raises a SystemError if it's not
present. We'll need to fix this before the JIT is really a viable way to
run things.

Original issue reported on code.google.com by jyass...@gmail.com on 24 Mar 2009 at 7:34

[GoogleCodeExporter]

Original comment by collinw on 29 May 2009 at 4:11

[GoogleCodeExporter]

Is this still relevant? Do we want to change it to be something more along the 
lines
of possibly persisting LLVM IR to disk? (This seems to be mostly about the
SystemError we used to raise.)

Original comment by collinw on 30 May 2009 at 1:47

[GoogleCodeExporter]

Yep, now it's just a speed optimization, to allow us to avoid compilation and 
optimization overhead on subsequent runs.

Original comment by jyass...@gmail.com on 30 May 2009 at 2:15

• Changed title: Persist LLVM IR to .pyc files
• Added labels: Type-Enhancement
• Removed labels: Type-Defect