[DGD]object management

[Par Winzell]

Hi,

 > Let's say,
 > 
 > 1. fork     A->B->Cx
 > 2. fork     A->D->E->Fx
 > 
 > with -> : inherited by, and Cx/Fx are leaf objects (may have clones).

 > But what happens if the upgrade on the first fork works, thus creating
 > a new issue of A, while the one on the second fails, leaving the old
 > issue behind? So how do I ensure that multiple issues don't exist?

I'm not sure precisely what Felix wrote, but you are right; until we
have full atomic functionality (and probably not even then for large
systems) you can have many issues existing at once.

This happens to me just about daily; I change the interface of some
core class, I upgrade it, I get the leaf-recompilation errors due to
the interface change, and I go through the leafs one by one and make
them work right. When the last one properly recompiles, the obsolete
issue is finally freed.

So... you can't ensure it, nor do I think you want to. Keep track of
issues using the status() index field and you'll have a unique key.

Pär

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