[Insight-developers] hash_map

[Miller, James V (CRD)]

I think this website has an implementation of STL that has a hash_map.  We might be able to just grab
this implementation since it is already an STL implementation.

http://www.stlport.org/

But like the SGI, its implementation of hash_map may depend on the rest of its implementation of STL.


-----Original Message-----
From: Brad King [mailto:brad.king@kitware.com]
Sent: Tuesday, February 27, 2001 10:24 AM
To: Insight Developers
Subject: [Insight-developers] hash_map


Hello all,

We at Kitware have discussed various approaches to a hash table, and have
decided that it would be best to add an STL hash_map implementation to
Insight's repository.  Our problem is that the SGI implementation of the
hash_map header is interdependent on most of the rest of SGI's stl
implementation, and cannot be included alone without a large amount of
hacking.  If someone would like to volunteer to take the time to do this,
the job is yours.

I'm asking the list for opinions and arguments about a hash table
implementation.  Is it worth adding one now?  Does WatershedImageFilter
depend heavily on a hash table, or would the stl map suffice?

There are many design decisions to make in writing a generic hash
table.  Does it automatically expand the bucket count when the load factor
is too high?  In what order do iterators move across the buckets and their
contents?  What is the implementation in each bucket, an stl list?

I'd like to get some discussion started about this so we can determine now
what is best for the project.

-Brad


_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers