[DGD]compile_object() kfun

[Erwin Harte]

On Tue, Mar 13, 2001 at 11:43:49PM -0500, Stephen Schmidt wrote:
> On Tue, 13 Mar 2001, Par Winzell wrote:
> >  > Design question: Should the update command call create()...?
> > 
> > Hell, no. That'd kill 98% of the reason for upgrades in the first place;
> 
> Yah, that occurred to me about 5 minutes after I sent
> the post off. Murphy's Law in action :(

Ah, my good friend, tell him I said hi. :)

> Perhaps as a flag on the update command, update -c or something
> like that to invoke create after the update? Or perhaps a
> separate command that would call create()?

Separate.  Call it 'recreate()', 'patch()', 'adapt()', whatever, but
keep it separate, definitely.

The create() function should contain the code to initialize a brand
spanking new fresh object, adding this to its burden is not good.

> > typically for a daemon object the create functions reads something like,
>  
> > 	static void create() {
> > 		user_objects = ([ ]);
> > 	}
> > 
> > and when you make a small API modification in your sixth year of running
> > a successful commercial game and your 'update' command deliberately zaps
> > your user database, as sort of a side-effect of recompiling the code,
> > you're going to be pretty damn pissed. :-)
>  
> But: Surely you would be saving data to a file, and create()
> would reload from the file? Or is the persistence so good
> that saving to file is no longer needed?

The latter.  Definitely the latter.  Once you've gotten the hang of it
you never want to change back anymore.  :)

Erwin.
-- 
Erwin Harte      : `Don't mind Erwin, he gets crabby. :)'
harte at xs4all.nl  :    -- Par Winzell <zell at skotos.net>

List config page:  http://list.imaginary.com/mailman/listinfo/dgd