[MUD-Dev] Database vs. Disk

J C Lawrence claw at kanga.nu
Fri May 23 01:07:21 CEST 2003


On Thu, 22 May 2003 16:06:37 -0400 (EDT) 
Zach Collins <(Siege)" <zcollins at seidata.com>> wrote:
> On Thu, 22 May 2003, Nicolai Hansen wrote:

>> Despite my search I have never encountered a database that allowed me
>> to store an object. I always have to write Object.Save() /
>> Object.Load() methods which is a pain, especially when you want to
>> change stuff.

> I believe a number of interpreted languages offer object serialization
> - Python and Java at the very least...

cf ZODB for the Python world.

> ... - but compiled languages like C++ tend not to have options like
> that by default.  

Certainly there are a great many external libraries which offer object
(de-)serialisation for C++ (usually in platform dependent ways), but
yes, its not a base language feature.

> In part, I think, this is because objects are more complex after
> compile-time than the sum of their parts as coded, so that you have to
> pick and choose what gets serialized and how.

Quite.

  Given: ObjectA contains a pointer to ObjectB, a reference to ObjectC,
  and an attribute of ObjectD, and inherits from ObjectX and ObjectY,
  each of which potentially have their own variously complex multiple
  inheritance trees.

When I serialise ObjectA, how much do I have to serialise? For that
choice, how do I guarantee logical consistency across the DB for
crash-recovery or rollbacks or even simple updates to objects which are
commonly inherited from or otherwise referenced?

--
J C Lawrence                
---------(*)                Satan, oscillate my metallic sonatas. 
claw at kanga.nu               He lived as a devil, eh?		  
http://www.kanga.nu/~claw/  Evil is a name of a foeman, as I live.

_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list