BS>> I guess it does the same to pack the index structure!

 WM> Not really -- my way is portable; "__attribute__ ((packed))" is
 WM> gcc- specific. My way also eliminates little-endian dependencies,
 WM> and is compatible with 64-bit systems with 8-byte longs. I realize
 WM> that the rest of Maximus isn't clean that way yet, but one has to
 WM> start somewhere. ;-) 

 BS> You have a point!

He has a point, Bo, but the '__attribute__ ((packed))' setting is what Wes
choose to get things working.  We should keep the code that way for now for
consistancy.  Since no one has stepped up to help with getting the code
running on other systems using anything other than GCC, I'm not as
concerned at this point.  At this point, the only systems not likely to use
GCC are ones where what Wes has done will work.  And the only one currently
wanting to use the code on big endian or 64 bit platforms right now is
Wes....  Now, if some one else steps up to help in this effort, supporting
with compiling and running in some of the different environments that you
and I currently don't have as easy avilablity to, we could work out an
agreement on how to change things....  [If we got set up properly for using
the sourceforge compile farm, we could do more testing of 64 bit and big
endian issues....  But I believe that is still using GCC for a
compiler....]

On the other hand, both Wes and I do want to get the code portable between
big and little endian issues.  I haven't looked up what Linux has in their
libraries for this.  I am familiar with NeXTStep's / Open Step's setup.  If
there is some "standard" way the Linux code is being cleaned up
for endian issues, then we should consider that method first....  I'll
admit, I haven't investigated this issue yet...  There are other areas of
the code still needing clean up just to get it running on an Intel CPU
based linux platform....

Take care.....

Bob Jones, 1:343/41


--- Maximus/2 3.01