3 ms·
"Spur is just a stepping stone to a more ambitious goals planned over the next five years." Does anybody know of a document describing these goals?
by asb 11y ago
"Spur is just a stepping stone to a more ambitious goals planned over the next five years." Does anybody know of a document describing these goals?
- lmz 11y agoThese ones maybe: http://www.mirandabanda.org/cogblog/cog-projects/ http://www.mirandabanda.org/cogblog/cog-projects/ (linked from the Release Notes: http://wiki.squeak.org/squeak/6207 http://wiki.squeak.org/squeak/6207 )
- gecko 11y agoAnd just to give the quick summary, Spur enables: * More immediate types. The classic Squeak/Pharo image type only allows for objects, strings, and integers as immediates, if I recall correctly. Spur adds individual characters, floats (soon), and some other stuff I'm spacing. Since all Smalltalk values are nominally heap-based, having more immediates is a really big deal. * Allows for a segmented heap. For desktop apps, this isn't a big deal, but for server apps, it means that the VM can actually relinquish RAM. * Better object pinning, which makes FFI a lot easier. * A more GC-friendly design, that should enable a much faster GC pass. The only downside, IIRC, is that the total image size grows by about 15%, but that's a pretty good trade-off, given a normal image is usually in the tens of megabytes.
- needusername 11y agoThreaded FFI has been on that list for a long, long time as well.