[ANN] Chuck type inferencer
cputney at wiresong.ca
Sat Jun 5 18:22:44 UTC 2004
On Jun 5, 2004, at 11:45 AM, Marcus Denker wrote:
> Yes, but that would be totaly transparent for the user (e.g. the
> developer). A dynamic optimizing system always de-optimizes as soon
> as you try to look harder (e.g with the debugger).
> So all the magic of AOStA is pretty much invisible. On the other hand:
> Having a Jitter around by default means that we don't care
> about how bad the non-optimized code is wrt. to execution. We could
> use the AST directly. (Or better, a compressed version like
> franz' Slim Binaries...).
Yes, I've been thinking about this as well. There's an interesting
paper on compressing syntax trees using the abstract grammar as a
statistical model, which I'm going to implement at some point.
<http://www.ics.uci.edu/~cstork/ire2001.pdf>. It's designed so that
most of the work is done by the compressor - decompression is fast.
The first step would be to create a binary format for distributing
Squeak code that would be faster to load than fileOuts - ie, decompress
the AST and generate byte code without having to do any parsing. Beyond
that would be modifying the interpreter to execute the binary format
> So "CompiledMethods are flattened parsetrees" does not mean that you
> would use the cm we have now as the AST, but use the AST as the
> compiled methods...
Yes, "Parse trees are structured CompiledMethods". I'd love to see
Squeak executing ASTs with the hotspots optimized by Exupery.
More information about the Squeak-dev