Re: Maven 4.0.0 namespace from an XML perspective

Le 04/04/2026 à 14:39, Romain Manni-Bucau a écrit :

> Im not sure what you mean by not xml parsers are low
>
In this debate, when we raised the argument that changing the namespace 
makes more difficult to use standard XML tools, it has been objected 
that no-one should use standard tools anyway and that POM should be read 
only through Maven API. I disagree, but this point is nevertheless 
important because this namespace change debate indeed become a small 
detail if the POM files are not intended to be read by standard 
namespace-aware tools.


> For model version v4.0.0/v4.1.0, ignoring the namespace but capturing 
> it to be able to have the info is the easiest to handle it - at least 
> with jaxb.
>
Configuring JAXB for ignoring the namespace is indeed a workaround, 
similar to configuring `javac` for ignoring warnings. It just hides the 
problem.


> I know xslt is referenced a lot but that is just a way to map models
>
The email that Chris has sent a few hours ago was not about using XSLT 
for mapping models. He was using the way that XSLT schema is declared as 
an example of how different versions of a XML schema are managed by the 
organisation who defines XML.


> a superset of both models is way easier to maintain than a mapping 
> with a third pivot format (for now).
>
Yes I agree: what you just described is option B. With option B, the 
Maven 4 POM is a superset of the Maven 3 POM. But with option A, the two 
POMs are completely unrelated from a XML point of view: none of them is 
a superset of the other, which is the reason why option A is more 
difficult than option B to handle using standard namespace-aware XML 
tools. Not impossible to handle, but more difficult.

     Martin