Optimizing stream operations

Tagged:  •    •    •    •    •  

One of the improvements in GNU Smalltalk 3.1 was the optimization of block operations on Streams. Basically, methods such as #next: and #nextPutAll: were rewritten to use fast collection operations instead of copying data element by element to the stream. Since #replaceFrom:to:with:startingAt: (the basic Smalltalk "move" operation) ultimately boils down to a memmove call, this is extremely fast.

But was this the end of the story? Today, after getting some inspiration from Nicolas Petton, I decided to try profiling Seaside and Swazoo.

Syndicate content

User login