A string manipulation benchmark has been written in lfyre and compared with the _same_ code written in C++.
Depending on the architecture, the lfyre code resulted to be 70% - 100% faster, i.e. ran at about double speed than C++ code.
A reference benchmark using standard C++ STL strings to perform the same operations resulted to run from 5 to 10 times slower than the lfyre code.
These results offer a first validation of the approach used by lfyre compiler to optimize temporary objects returned by functions.
All the three benchmarks are available in lfyre sources, in the directory examples/, to allow public review and execution.
The bootstrap compiler, lfyre-level0, is now minimal but usable.
It can translate simple lfyre code into C source code.
Only a limited subset of lfyre language features are currently supported by this compiler, yet they are enough to self-host further compiler development, by writing source code in lfyre instead of C.
The lfyre project has been just registered to SourceForge.
For the moment, use the mailing list lfyre-devel@lists.sourceforge.net
to participate to lfyre technical discussion and development.
Forums and Trackers will be used when the lfyre project
reaches a sufficiently high activity rate.