On Monday 23 Oct 2006 11:49 am, David Matthews wrote:
Thanks to some financial support from the Verisoft project and TUM I've been working through the summer on an updated version of the Poly/ML run-time system and a new code-generator for the AMD 64-bit architecture.
The major new features are: Support for stand-alone binaries Support for additional platforms: AMD64, Intel Macs, Cygwin No artificial limits on size of heaps or saved image Uses standard GNU tools for building Fixed address mmap and trap-handling removed
This comes at one significant cost: the persistent store system that has been a feature of Poly/ML almost since the beginning has been removed. The reason for this is that to be effective the persistent store needed to be able to reload a database at the same address every time but that could conflict with other libraries and result in seg faults. Rather than constantly adjust the location of the database with each new kernel release I decided to make the leap and change to making stand-alone binaries. This achieves much the same as the persistent store but without the complications.
One advantage of the persistent object store is that it is organised as a hierarchy: if many people want to share a body of ML code and data, the common material can be put in one database and the individuals can work in descendants of that database. At least one group of ProofPower users have large applications that are only feasible when organised in this way. Is anything like this possible under the new set-up?
Regards,
Rob.