Hardcoded header depth

[opoudjis]

https://github.com/metanorma/metanorma-plugin-lutaml/blob/4ffdcaad5c1ac6a6c0bf3f1ac6f41dd26639fa7e/lib/metanorma/plugin/lutaml/lutaml_uml_attributes_table_preprocessor.rb

def table_template
          <<~TEMPLATE
          === {{ definition.name }}
          {{ definition.definition }}

This means that any instance of lutaml_uml_attributes_table ever will always result in a section number one level deep.

It should go without saying that this is unacceptable: I should be able to get lutaml to generate clauses at any level depth I choose, in general.

In fact, I think it is completely wrong for Lutaml macros to be generating the title of the section at all. If it must do that, you have to give me the option of choosing the section depth.

[opoudjis]

This has been partially addressed, by 1dbd31a : titles are now optional, and can be disabled with the "skip" parameter.

But now that lutaml is importing correctly, things are worse than that. The IDs automatically being generated for titles in lutaml_uml_attributes_table instances is colliding with anchors present elsewhere in the document. This results in invalid Metanorma documents.

For example,

<clause id="term" inline-header="false" obligation="normative">
<title>Term</title>
<figure id="_ff58c387-3ea4-4388-ad17-fa0bc801e3a1">
<name>Terms and Definitions elements of StandardDocument</name>

is being generated by lutaml_uml_attributes_table, and it collides with a instance of term already defined as a clause ID elsewhere.

Metanorma determines ID assignment for assets. Without exception. It is responsible for their uniqueness. If this functionality is to persist, I am going to overwrite any anchor generated by lutaml with a GUID, to avoid collisions, and I'd rather lutaml do that.

As is, I am removing every single title in lutaml_uml_attributes_table instances in the document I am working on, and replacing them with skip.

[ronaldtse]

I should be able to get lutaml to generate clauses at any level depth I choose, in general.

Agree.

The IDs automatically being generated for titles in lutaml_uml_attributes_table instances is colliding with anchors present elsewhere in the document

Isn't this easily solved by having lutaml change the IDs generated? e.g. lutaml_xxx?

Why is skip even necessary?

[opoudjis]

Why is skip even necessary?

To have the title hardcoded in the YAML is absurd. You have no business telling the author there shall be no content between the clause title and the generated table: only the table is automated content; the title I put it under in my document is my business.

And you certainly cannot hardcode the depth of the title, as was the case: that is just absurd.

[w00lf]

@opoudjis @ronaldtse what should we do with entity name(definition.name)? Currently, its rendered like that:

=== {{ definition.name }}
{{ definition.definition }}

{% if definition.attributes %}
{% if definition.keyword == 'enumeration' %}
.{{ definition.name }} values

[ronaldtse]

@w00lf we can either use the AsciiDoc include to increase the depth, or use a Liquid counter to determine the number of = signs, as we have done before.

[w00lf]

use the AsciiDoc include to increase the depth

@ronaldtse do have an example?

[w00lf]

In case we will be using nesting level variable we will need user to pass depth each time, is it ok? @opoudjis

[ronaldtse]

I think it's something like:

include::myfile.adoc[level=+1]

[ronaldtse]

In case we will be using nesting level variable we will need user to pass depth each time, is it ok? @opoudjis

You can set a default level if user didn't set it.

[w00lf]

In case we will be using nesting level variable we will need user to pass depth each time, is it ok? @opoudjis

You can set a default level if user didn't set it.

@opoudjis is it ok if we done it like that?

[opoudjis]

You can do that, sure. I'm not going to use it anyway, because I feel the title should be decided by the author.

[ronaldtse]

@opoudjis how do you plan to "decide" the title for an enumerated tree of packages? Providing a pattern that accepts the name of the package/class? Can you elaborate how that can be possible?

[w00lf]

@opoudjis how do you plan to "decide" the title for an enumerated tree of packages? Providing a pattern that accepts the name of the package/class? Can you elaborate how that can be possible?

@ronaldtse lutaml_uml_attributes_table macro always accepts entity name to render, there will never be a package tree here.

[opoudjis]

@w00lf @ronaldtse Is this ticket still open?

[w00lf]

ping @ronaldtse