[prev in list] [next in list] [prev in thread] [next in thread] 

List:       mdbtools-dev
Subject:    Re: [mdb-dev] Data pages
From:       knyberg () grebyn ! com
Date:       2000-03-16 13:48:51
[Download RAW message or body]

My rule on this is as follows (it's known as Nyberg's law - feel free to
reference it!):

	  It's always easier to make a working program fast than it is to
	  make a fast program work.

Meaning, of course, take the inefficient, but (not in the formal sense)
correct, solution, and let your knowledge and understanding of the problem,
your programming skills, and the speed of computer hardware have a chance to
catch up.  Once you see the whole puzzle, it's often easier to figure out
where (and in what order) you should have put the pieces.

I have many times coded inefficient (but obvious) solutions to problems and
not gotten the elegant solution until much later.  Having A solution permits
me to move forward with other aspects of a system development without
waiting on the perfect solution or being completely blocked.

-- Karl --

   Date: Thu, 16 Mar 2000 08:34:08 -0500 (EST)
   From: Brian Bruns <camber@umcc.ais.org>
   List-Id: MDB Tools development <mdbtools-dev.lists.sourceforge.net>

   I just had a revelation about how the datapages work, and thought maybe
   someone would have a better idea.

   Table definition pages start out with '02 01' as the first two bytes, while
   datapages (and this includes catalog pages btw) start out '01 01'.  The next
   two bytes are currently of unknow use, but bytes 4-7 are the parent page of
   this datapage.  So our catalog pages all have '02 00 00 00' for a parent
   page (page 2...little endian remember).

   So, page 2 is some sort of definition for the catalog...not all that
   important to us at the moment.  

   Anyway, one approach to getting the datapages reading correctly would be to do
   the same thing we do for catalog pages, which is read the entire database
   for pages with a parent page of the table definition. This is grossly
   ineffiecient, however it would give the correct results until we can can
   find someway to get the proper page linkage (if there is a way).

   What do you think?

   Brian

[prev in list] [next in list] [prev in thread] [next in thread]