Safe data definition for library configurations

[Markus Elfring]

A RELAX NG schema is provided. I find that it contains still open issues. Would you like to improve it any further?

The pattern “[a-zA-Z_][a-zA-Z_0-9:,]*” for the item “DATA-EXTNAME” is inappropriate to express a list of valid function names (or aliases) because it is too generic.
Are there any more properties to consider?

[versat]

What exactly do you mean with "too generic"? Do you have any concrete suggestions?
I already stumbled over DATA-EXTNAME because it constains "," when changing platformtype nameentries (see https://trac.cppcheck.net/ticket/8401).
So i would agree to say it does not fit everywhere 100% at the moment.

[Markus Elfring]

What exactly do you mean with "too generic"?

• Should name space identifiers be separated by double colons?
• How should this item be validated if a function list was stored there?
• Would you dare to avoid the comma as a delimiter?

---

How are the chances to increase the development attention for safer data models and their validation?

[versat]

Should name space identifiers be separated by double colons?

Why not? How else?

How should this item be validated if a function list was stored there?

Not sure what exactly you mean with "this item".

Would you dare to avoid the comma as a delimiter?

Is there a problem with comma as delimiter?

[Markus Elfring]

Why not?

We are used to the specification of double colons in C++ source code.
But how does the current regular expression in the RNG file fit to this usage?

How else?

I suggest to extend the mentioned data model description.

Not sure what exactly you mean with "this item".

Which wording would you prefer for “DATA-EXTNAME”?

Is there a problem with comma as delimiter?

I imagine that delimiting functionality should be expressed by safer means in the XML schema.
Will a semantic data structure definition be more appropriate?