Separate documentation parsing and rendering #2081
Conversation
- Move Documentation out of markdown_processor and into model package - Create a renderer class for Documentation
|
|
||
| class DocumentationRendererHtml extends DocumentationRenderer { | ||
| @override | ||
| Tuple2<String, String> render(List<md.Node> nodes, bool processFullDocs) { |
There was a problem hiding this comment.
The old code had an optimization that avoided calculating links for parts of the documentation that would never be rendered (if it is only used in a oneLineDoc). This used to make a very noticeable difference on dartdoc's performance. It looks as though that optimization has been removed.
If it doesn't matter anymore we can probably drop the flag.
There was a problem hiding this comment.
If I understand correctly, I think we are still honoring that flag in MarkdownDocument.parseMarkdownText, it stops collecting markdown nodes if the flag is false. It's just a little harder to follow now because half the code path is there and the other half moved to the renderer. Documentation._renderDocumentation calls both methods in succession, passing the same value for the flag, so it should still be doing this optimization.
Move Documentation out of
markdown_processorand into model packageCreate a renderer class for Documentation