Re: [forces] draft-ietf-forces-model-extension

[Julian Reschke <[e-mail]>]

On 2014-10-22 14:54, Haleplidis Evangelos wrote:
> Greetings Julian,
>
> I'm a little confused on why you think there is no backwards compatibility.
>
> LFB definitions that use the 1.0 namespace are valid in the 1.1 namespace.

Unless I'm missing something: no, they are not.

Can you provide a 1.0 example, and show how it is valid according to the 
1.1 schema?

> LFB definitions that use the 1.1 namespace are not always valid with the 1.0
> namespace.

They are never, because they use elements from a different namespace.

> That means that a software that understands LFBs in the 1.1 namespace can,
> without any change, understand older LFBs. That was the end goal.

No. Unless I'm missing something it'll have to support both namespaces, 
thus both schemas.

> Software that understand the 1.0 namespace (even if we didn't change the
> namespace) will, of course, have trouble parsing the new xml if the
> extensions are used.

...well, unless you have a sane extensibility model to start with.

> This document did not change any LFB definitions developed with the 1.0
> namespace. As there was no reason for that as 1.1 is backwards compatible
> with 1.0
> The namespace is a way to distinguish the different versions and notify that
> in the 1.1 namespace these extensions apply.

Changing the XML namespace name essentially means you have a new 
vocabulary. No existing code (unless it was buggy enough to ignore 
namespace declarations) will recognize elements from the new namespace.

> Thus, I don't understand why old software will be needed to change in order
> to understand the old format.
> For the new format, of course you'll have to change it. But for the new
> format you'd definitely had to update the software if you want to be
> compatible with the 1.1 namespace, so I don't see the issue.
>
> Is something I miss?

At least one of us does :-)

Best regards, Julian