Your change to tags.rb is having a negative side effect. The regression test (just run "make") fails because some of the adoc formatting information is being removed. I want to see this so I can create the HTML equivalent.

11,13c11,13
<     "norm:superscript": "xyz 2^32^ 123",
<     "norm:subscript": "xyz X~i~ 123",
---
>     "norm:superscript": "xyz 232 123",
>     "norm:subscript": "xyz Xi 123",

I also added this diagnostic puts to your tags.rb and added another case to test.adoc to check macro replacement in paragraph (i.e. [[foo]]) anchors. Looks like the macros aren't replaced in paragraph tags.

tags.rb change:

        if node.text.nil?
          nil
        else
          out = node.apply_subs(node.text, node.parent.instance_variable_get(:@subs))
          puts("XXX: in='#{node.text}' out='#{out}'") if out != node.text
          out
        end

test.adoc change:

Original

Macro call [#norm:macro_call]#200 {ge} 100# outside.

Change

Inline macro call [#norm:inline_macro_call]#200 {ge} 100# outside.

Paragraph macro call [#norm:paragraph_macro_call]#10 {ge} 10# outside.

Output from diagnostic puts:

XXX: in='xyz 2^32^ 123' out='xyz 232 123'
XXX: in='xyz X~i~ 123' out='xyz Xi 123'
XXX: in='ABC [.underline]#inside tag# GHI' out='ABC inside tag GHI'
XXX: in='200 {ge} 100' out='200 ≥ 100'

You can see:

• Subscript & superscript formatting is being removed (bad)
• Underline formatting is being removed (bad)
• "ge" macro is being expanded in inline tag (good)
• "ge" macro is NOT being expanded in the paragraph tag (bad)

Interesting, thanks for testing - I'll take another look.

Any idea how?

Instead of calling node.apply_subs(node.text, node.parent.instance_variable_get(:@subs)), which passes in an array of substitution functions from the parent, could you just pass in a reference to just the macro substitution function?

If you omit the subs parameter to apply_subs(), it defaults to NORMAL_SUBS (see screenshot below). The NORMAL_SUBS is an array of function references and includes "macros" but also others we don't want. So, could you just pass in a reference to the "macros" function?

BTW, even if you can just get macro substitution to work, its only working in inline tags ([#foo]#...# syntax). How can we get it to work for the paragraph tags ([[foo]]) too?

@Timmmm, any update on this?

Hey, sorry - slightly fell off my radar. I'll take another look.

@Timmmm, I see that the upstream ISA manual now has macros everywhere instead of using Unicode characters or Unicode codes. Looks like you might have been working with @aswaterman on the change. This raises the priority of getting this PR fixed since now the normative rules won't show the intended Unicode characters (just the macro name).

I've merged in the upstream ISA manual repo and resolved about a dozen conflicting files. However, our normative rules are now somewhat broken due to the use of the macros. Can we get this fixed before someone notices and complains?

Ok it seems the loss of bold/italic etc was a separate issue to substitutions. Those things get converted to inline_quoted nodes and we weren't outputting any formatting characters for them based on their type.

This seems to work now but maybe double check!

I'm not clear from your comments about bold & italics. Can we add those?

Yeah they should work.

Ok I think all the feedback is addressed now.