Re: [Gnu-arch-users] Status of global and tree aliases

[Tom Lord]

    > From: Aaron Bentley <address@hidden>

    > Tom Lord wrote:
    > >     > From: Andrew Suffield <address@hidden>
    > > 
    > >     > No, these really are tree aliases. 
    > > 
    > > Did I misread?  Jblack gave them a "source" filename.  They would be
    > > archived.  In effect, they are persistently set (though mutable)
    > > per-version.

    > They're not per-version, they're per-tree.  

Yes, I realized that after I posted.

Still, I don't think that that's a semantic difference.  Rather, I
think that "per-version" and "per-tree" are equivalently expressive 
but optimize the syntax and implementation for different patterns of
use.

Basically, one way or another, we wind up with potential variable
definitions scattered in various locations, some persistent, some
precious within a project tree, some per user, perhaps even some
per-user-per-package and/or some per-working-dir-per-package.   We'll
need rules and mechanism to combine all these definitions in
controllable ways to get the right settings for various contexts.

    > > Actually, it would be undesirable if a merge could _not_ change a
    > > version variable.   It's only important that verges can change version
    > > variables in controlled and domain-specific ways.

    > The obvious ways of changing a version-variable by committing to a 
    > different version look quite heinous.  You'd have to either add a bogus 
    > patchlog for the target version, or alter an existing patchlog for the 
    > target version.  So I assume you have something more sophisticated in 
mind.

I'm thinking that there are some variable definitions whose value is a
set of variable definitions (basically, an xl program computing an xl
program as its result).  That plus some rules about when those
"computed definitions" are installed to become actual definitions.

-t