navit/fib-1.1/README

Version 1.1 now supports increasing the key using the fh_replace*
functions.  Previously it would simply return NULL when you tried to
increase the key.  It also improves performance slightly by only calling
checkcons when we are about to use it, at extract, instead of calling it
on every insert.

I have now fixed fh_union and it properly updates the minimum.

Thanks to Ryan Earl for pointing out that in fh_consolidate, it is VERY
time consuming to constantly malloc/free an array of pointers.  The array
is small enough that simply reallocating when more pointers are needed
is ok.

Thanks to Thomas Eschbach and Wolfgang Guenther who have pointed out bugs
with my code.  Wolfgang Guenther provided a fix which put in on the correct
track for where the bug was.  They have also provided a few test programs
that exhibited other bugs which I have now integrated into my source so
that you can easily regress test the library.

I have reciently completed a review of the code.  I have made a number
of improvements with a few minor interface changes.  There is another
improvement to the rh_replace* family of functions to help eliminate
redundant code.

I'm still planning on writing a type safe memory allocator for use with
the code instead of using malloc to hopefully improve performance slightly.
1	Version 1.1 now supports increasing the key using the fh_replace*
2	functions. Previously it would simply return NULL when you tried to
3	increase the key. It also improves performance slightly by only calling
4	checkcons when we are about to use it, at extract, instead of calling it
5	on every insert.
6
7	I have now fixed fh_union and it properly updates the minimum.
8
9	Thanks to Ryan Earl for pointing out that in fh_consolidate, it is VERY
10	time consuming to constantly malloc/free an array of pointers. The array
11	is small enough that simply reallocating when more pointers are needed
12	is ok.
13
14	Thanks to Thomas Eschbach and Wolfgang Guenther who have pointed out bugs
15	with my code. Wolfgang Guenther provided a fix which put in on the correct
16	track for where the bug was. They have also provided a few test programs
17	that exhibited other bugs which I have now integrated into my source so
18	that you can easily regress test the library.
19
20	I have reciently completed a review of the code. I have made a number
21	of improvements with a few minor interface changes. There is another
22	improvement to the rh_replace* family of functions to help eliminate
23	redundant code.
24
25	I'm still planning on writing a type safe memory allocator for use with
26	the code instead of using malloc to hopefully improve performance slightly.