Hi Gary,
I think someone famous quoted:
Ours is not to reason why,
Ours to work around, or die.
Yes, it does look a meaty bit of code.
From looking at Insert(), it only tries increasing the Hashtable
if it's reached the limit (specified by the capacity (number of
buckets) and load factor) which is kept in the private variable
loadsize.
Did you try my suggestion of seeing how the Hastable grows by
using Add Watch?
The code I used is:
Dim oHashtable As New Hashtable
For I = 1 to 50
oHashtable(i) = 1000 + i
Next
You'll see that loadsize is small (7 for me). When the 8th item is
to be added, Insert() finds loadsize exceeded and proceeds to expand
the hashtable. In the docs it says that "the number of buckets is
automatically increased to the smallest prime number that is larger
than twice the current number of buckets.".
The initial (for me) number of buckets is 11 and the loadfactor is
..72 which gives a limit (loadsize = buckets.Length * loadfactor) of 7
(7.92 rounded down).
So, double the number of buckets (22) and go to the next prime
giving a new number of buckets, 27. The loadfactor stays the same, so
the new capacity (loadsize) is 27 * .72 = 24. The 25th insert will
cause further expansion. And so on.
So, your job is to find an initial size for the Hashtable which
will not cause Insert to expand the table. If it doesn't expand, it
won't reach the Exception.
I don't know how many items you have. But whatever it is, you need
to calculate for a loadsize which greater than your number of items.
Let's say you have 51 items. The loadfactor is .72 so the number
of buckets will be at least 51 / 0.72 which gives 70.833 which this
time is riunded <up> to 71. The number of buckets must be a prime
number (for an effective hashing algorithm). 71 <is> a prime number so
that's no problem.
Let's say you have 52 items. The loadfactor is .72 so the number
of buckets will be at least 52 / 0.72 which gives 72.222 which this
time is rounded up to 73. The number of buckets must be a prime number
and the next prime is 89. So the table will have 89 entries.
Have you followed so far? Because now I'm going to say forget all
that, lol, that's just background.
For plain Hashtables, the constructors are (), which you've been
using, (iCapacity) and (iCapacity, fLoadFactor).
You can use either of the other two because all the calculations
above are done by the Hashtable. It will ensure that whatever
loadfactor is used (default or supplied) it will have a sufficiently
large table to handle the given number of items.
The loadfactor is effectively the slider between max memory
efficiency (loadfactor = 1) and max access efficiency (loadfactor
smallish).
Small hashtables (loadfactor high) will have lots of what are
called 'collisions' when the initial index of two or more items is the
same and steps must be taken to get round this. With large hashtables
(loadfactor low), an item's initial index is always available to just
that item. But for this to happen there will be a large number of
buckets that will never be used. With medium Hastables there will be
some or occasional collisions, depending on how 'medium' it is.
Still with me ? Because now I'm going to say forget all that
<again>, lol.
I wouldn't worry about the loadfactor (should have said that in my
last post) - just use the capacity.
Hope this gives you more than you need. :-).
Regards,
Fergus
