Architectural issues regarding subparsers in ctags.

[EvanCarroll]

Tagbar does not currently support ctag subparsers (as implemented here universal-ctags/ctags#2070) That is it determines the filetype and spawns off to ctags by forcing --language-force= and hard coding the kinds with --$lang-kinds=kinds. You can see this here

• https://github.com/majutsushi/tagbar/blob/387bbadda98e1376ff3871aa461b1f0abd4ece70/autoload/tagbar.vim#L1121
• https://github.com/majutsushi/tagbar/blob/387bbadda98e1376ff3871aa461b1f0abd4ece70/autoload/tagbar.vim#L1122

What needs to happen is,

• These two lines above need to be purged in tagbar.vim
• We need let ctags figure out the language
• We need to let ctags provide us all the kinds available, because we can't enumerate types that may or may not be present in subparsers in a configuration for the parent parser.
• We need to remove the ctagstype and kind properties from all languages in uctags.vim
• Instead of placing the names of kinds to resolve the single letter to the display name we should use --output-format=xref which provides the resolutions for us. This means if a subparser is invoked we can use the output to decide the display name, rather than having to code that in our own configuration file.
• We need to delete the hard-coded clearing of --extra because that stops subparsers https://github.com/majutsushi/tagbar/blob/387bbadda98e1376ff3871aa461b1f0abd4ece70/autoload/tagbar.vim#L1094

If you want me to do the work for uctags, I'm happy to help out.

[EvanCarroll]

One of things we may want to do is simply only support universal ctags. If that's the case we can really sanitize this implementation easily. I'd even rather us self-package ctags in a ./.vim/tagbar/bin rather than fall back to etags. If that's ok, I can whip up a patch reduce the code as described above, and support subparsers.