Infrastructure for warnings

[philshafer]

Lots of non-fatal things can kill you.

Like having the wrong URI for a namespace. Your script looks fine, runs fine, but does not work.

Example:

ns xnm = "http://xml.juniper.net/xnm/1.1/xnm-oops!!";

...
    if (not(xnm:error)) {
       call destroy-the-world();
    }

Since xnm:error will never match, the world will be gone. Not good.

I think I can make this an optional warning, if that helps. The rule would go something like this:

When libslax sees a chunk of XML, it will looks for top-level 
namespaces whose prefixes match a prefix defined in a top-level
"ns" statement in the main script.

This will allow us to catch errors like your's but without searching the entire new XML document, which could be expensive.

I want to put this all inside a new "warning" infrastructure, so the user can say "-Wprefix-reuse" for this first scenario, -Wprefix-reuse-deep for complete namespace checking, -Wxxx for some future case, and "-Wall" checks these and other future warning flags.

The same logic can be applied to element names as well, so a script can report (after execution) any XML element names used in XPath expressions that were not seen during execution. Again, not check, but if you said "if (xnm:errror)", it would be a complete lifesaver.
Thanks,
Phil