[freeside-devel] svc_broadband.actypenum

ivan at 420.am ivan at 420.am
Mon Sep 23 05:22:45 PDT 2002


On Fri, Sep 20, 2002 at 10:14:15AM -0700, Kristian Hoffmann wrote:
> If you'd like, I can roll-back the merge of the svc_broadband branch
> because I'm not sure we'll have a way to fix this soon.

Well, it's still a development tree, so if you're going to continue to
work on it, no reason to roll it back.

On the other hand, if development on it is stalled, and no one
else is interested in finishing it...  I'm not sure what I'd do with it.

> I just don't see another way to prevent someone from setting up a service
> defined as "Wireless, Resideitial, 384Kbit" with a DSL access
> concentrator.

Not sure I understand you here.  If by "someone" you mean the sysadmin
configuring service definitions, I don't think safeguards to prevent
them from doing something are worth denormalizing the database.  There's
lots of ways you can do dumb things with service definitions (you can
define fixed values for username and _password!), I can't really say I
consider it a big problem.

>  I guess we're looking at svc_broadband as a meta-service
> using ac_types to seperate physically different instances of
> svc_broadband (DSL, Wireless, CoLo, etc.).  Am I making any sense?

Perhaps "ip allocation" should not be a svc_ item like svc_broadband,
just a tracked piece of data (like domain_record), and there should be
regular svc_ tables for DSL, Wireless and Colo...

-- 
_ivan


> 
> -Kristian
> 
> 
> On Fri, 20 Sep 2002 ivan at 420.am wrote:
> 
> > On Fri, Sep 20, 2002 at 07:54:22AM -0700, Mark Wells wrote:
> > >
> > >
> > > On Fri, 20 Sep 2002 ivan at 420.am wrote:
> > >
> > > > Kristian, Mark,
> > > >
> > > > Please remove svc_broadband.actypenum as it is breaks database
> > > > normalization.
> > > >
> > > > As the folks who fixed the terribly-bogus `owed' field I wouldn't have
> > > > expected you to be doing this sort of thing.  :)
> > >
> > > I'm sure we'll fix it once we find a way to generalize the fixed/default
> > > field mechanism to enforce weird constraint expressions or enumerated
> > > values instead of a single fixed value.
> >
> > I'll try to help out, but I'd appriciate it if you could take on
> > making sure this happens.
> >
> > >  Until then, it's necessary.  The
> > > service definition has to be able to mandate the selection of an ac with a
> > > particular ac_type.
> >
> > I'd sooner rip out ac_types entirely (they seem pretty extraneous) than
> > live for long with a denormalized database.  If you can, please make
> > sure this happens soon.  We can't release with this sort of thing.
> >
> > In the future, please don't check things onto the HEAD that put things
> > in an unreleasable state like this.
> >
> > --
> > _ivan
> >
> 



More information about the freeside-devel mailing list