Talking about GBs it makes me wonder , even if the image allowed more than 500MBs if that would be a wise move.
Smalltalk is afterall one of the slowest , execution wise, programming languages. In python which is as slow if not slower, the attitude is to move big data to C libs and then interface with these C libs via python. When I see even morphic at times take up to 50% of my dual core 2.0GHz CPU just makes me wonder if cramming 1 GB or more inside the image will end bringing pharo to a crawl for processing that data. So what pharo may need is something that already nativeboost is doing, mapping to C types and access to fast C functions for processing and manipulating big data. Just make it more easier. For example python has numpy (mainly C code, extra types etc) and nowdays also Blaze is emerging as a future candidate. https://github.com/ContinuumIO/blaze <https://github.com/ContinuumIO/blaze> just a thought. -- View this message in context: http://forum.world.st/Pharo-cog-Vm-on-Windows-with-2GB-virtual-mem-tp4710746p4712155.html Sent from the Pharo Smalltalk Developers mailing list archive at Nabble.com.