Re: [Vrs-development] CProgamming setup

[Chris Smith]

On Tuesday 16 April 2002 15:40, Bill Lance wrote:

> From what I've started with the Cluster Image design,
> it's looking like a hiearchial key tree structure.  In
> perl, I'd implement this as a hash-of-hashes.  But
> that can be a real bitch to serialize.  

It's not to bad if you think about it.  You just walk
the tree, stepping back up once you've done a leaf node.
In fact, I've got an ace bit of C code that does just
this...


> Since the pert part is intended to be temporary anyway,
> perhaps we should think about setting this up as a tie 
> to something external that C can see too.  Althought
> that might mean a disk file, which I really wanted to
> avoid for security reasons.

No need.  Besides, thats too complicated.
Whatever code is written to handle the Cluster Image
also provides methods to read and write entries to
the image.  The implementation details are hidden from
all other areas of the system.

If you want to get something out of the image, you
invoke the appropriate method (which is a GWService
call).

The parameters that are passed to methods (GWServices)
have to be consistent though, particularly if some
GWServices are perl and others are C/C++ or whatever!


The way you design/implement stuff using middleware
message calls will become much clearer when I release
the Domain'ed version of Goldwater and the BASE
architecture for the CM... something for everyone
to look at.

Coming soon.... gimme a week-10 days....

Chris.
-- 
Chris Smith
  Technical Architect - netFluid Technology Limited.
  "Internet Technologies, Distributed Systems and Tuxedo Consultancy"
  E: address@hidden  W: http://www.nfluid.co.uk