Is tree_cmd option ignored?

[vbauerster]

None of tree-* commands respect my config, which has some blacklist for go file types.
I've tree_cmd option set in my kakrc and it expands correctly:

set-option global tree_cmd "kak-tree -c %val{config}/kak-tree.toml"

This is content of my kak-tree.toml:

[filetype.go]
blacklist = ["nil", "block", "identifier"]

For example, if I issue tree-node-sexp, I still can see blacklisted items.

[ul]

tree-node-sexp ignores white/blacklist by design (we could parametrize it to support both modes though). Do you have problems with blacklist when using other commands?

[vbauerster]

Yes I do, for example :tree-select-children nil<ret> selects nil occurrences, despite it's blacklisted.
I would much appreciate if tree-node-sexp could drop blacklisted items, because it's too noisy by default. My workflow is first to run tree-node-sexp on selection in order to acquire definition node needed, then run :tree-select-* node_needed<ret>.

[vbauerster]

IMHO it's not necessary for tree-select-* commands family to obey filter rules (because their argument is filter in fact), but tree-node-sexp must obey in first place.

[vbauerster]

Also I wish tree-node-sexp had -lock switch, to freeze its menu.

I use kak-lsp as well with gopls which spawns to many unnecessary errors, so most of the times menu of tree-node-sexp gets suppressed by gopls.

[ul]

Yes, that sounds reasonable. Also, I think we should improve the way blacklisted nodes are handled in general, it might worth not only scanning for visible nodes but removing invisible nodes from the tree in the first place (and lifting up their visible descendants). It will make navigation result more intuitive.

On an unrelated note, you can reduce noise from kak-lsp by overriding lsp-show-error command to filter our harmless errors from gopls.