Longer term solution for extension name dictionaries

[bashbaug]

I don't think it would, unfortunately, because AFAICT the extension name without any suffix is defined as an attribute by the build system, in gen/specattribs.adoc, the attribute just doesn't have any value so it doesn't generate any text in the spec. Maybe we could simply update the Makefile to generate attributes with some text instead, say "OOPS!", and that way at least the incorrect attribute usage wouldn't be silent?

Would it be helpful to just define the attribute's value as its name, or an xref to the extension anchor in the appendix? Trivial to do.

Originally posted by @oddhack in #1131 (comment)

[bashbaug]

Figured it would be easier to track this discussion in an issue vs. a merged PR.

Would it be helpful to just define the attribute's value as its name, or an xref to the extension anchor in the appendix? Trivial to do.

I think this really depends on our goals:

1. If we want the attribute to generate passable text if it is used un-suffixed then something like this is the right thing to do.
2. If we want to identify and fix cases when the attribute is used un-suffixed then we want to make it as obviously wrong as we can.

I think either would be preferable over our current "silently generate no text" behavior.

I keep hoping there's a better overall long-term solution, though...

[oddhack]

IME no matter how much automation is thrown at checking markup - Vulkan and OpenXR have quite a few different scripts looking for different point errors - people can always find new ways to commit markup errors :-(

[oddhack]

(Related example, you might be surprised how many duplicated connectives like 'and and' / 'the the' are to be found in the spec source).