Thanks, I've increased this limit but I'm not sure how it's supposed to work then.

I assume it's an issue between nvim, the picker and the LSP.

If I runphpactor index:search --fqn-begins=SomeNamespace it works as expected and very quickly gives me the complete set of results.

However,Snacks.picker() and telescope both cause phpactor to go oom when runninglsp_workspace_symbols. Perhaps I need to create a custom picker implementation that callsphpactor index:search via CLI, though this feels like it should be possible with the existing implementation

edit: Seems that I had to setcore.min_memory_limit when loading the nvim plugin rather than.phpactor.json for using the plugin and it builds the index independently of the CLI.

Though now I'm getting

[ERROR][2025-05-06 14:37:30] ...p/_transport.lua:36     "rpc"   "phpactor"      "stderr"        "Starting language server, use -vvv for verbose output\n"[ERROR][2025-05-06 14:37:59] ...p/_transport.lua:36     "rpc"   "phpactor"      "stderr"        "\n"[ERROR][2025-05-06 14:37:59] ...p/_transport.lua:36     "rpc"   "phpactor"      "stderr"        'In LspMessageSerializer.php line 23:\n                                                                               \n  Could not encode JSON: "Malformed UTF-8 characters, possibly incorrectly en  \n  coded"                                                                       \n                                                                               \n\nlanguage-server [--address ADDRESS] [--no-loop]\n\n

I'll try to tun on-vvv mode from within the plugin

edit 2:

I don't think it works the way I expect it to work.

If the lsp workspace symbol limit is 250 and I open snacks.picker I expected that:

• it would show me 250 results (likely meaningless as there is no search term)
• as I type it queries the LSP and fetches max 250 relevant results

But it seems that what actually happens is that those 250 results are the only ones indexed. I think with a higher limit it's crashing the lsp because it's trying to return all ~60k files at once.

A 250 limit would be fine if the second step above worked and I could search the entire workspace but only return 250 relevant files