Finally, the new latest changes on the data-type engine get
committed. It took a lot longer than expected but everything is now a
lot easier to understand (ok this sentence is very relative), smaller
in number of functions and a lot faster. The idea was to make the
pack/unpack for sparse data-types as fast as possible but as a side
effect the convertor management get slim-fasted (Netpipe on SM show
improvements for all sizes in order of 2-5%).
I'll continue testing it but please help me. If you do not plan to
make performance tests then compile in debug mode as in this mode the
data-type protection mechanism will print a lot of information if it
detect that something bad happens (like writing outside the user
space buffer ...). If suddenly your program start dumping a lot of
text please copy it in the email where you report the bug, and it
will make the debugging process a lot faster for me.
Thanks,
george.
"Half of what I say is meaningless; but I say it so that the other
half may reach you"
Kahlil Gibran
