🐞 Bug: link-lmstudio does not work with custom LM Studio directory

[thanhbinh01234]

Description

When running gollama -link-lmstudio with -lm-dir flag, it still searchs default directory of LM Studio instead of specified directory.

How to reproduce

Run command gollama -link-lmstudio -lm-dir <custom-lm-dir> -ollama-dir <custom-ollama-dir>

Screenshots / Logs

Environment

• macOS version: 15.3 (24D60)

Can you contribute?

• I will attempt to implement a fix for this bug

How did you find out about Gollama?

Google search

[sammcj]

Howdy folks,

Can you please confirm if this is now resolved after merging #169

Thanks,

[erg]

Hey, I contributed some of the initial code for this patch. I tried to link my lm-studio on my other laptop and it didn't find the old default lm-studio directory on macOS.

• Old default: ~/.cache/lm-studio/
• The new default is ~/.lm-studio

Maybe it's beyond the scope, but if the user has the old default we could fix it with a symlink ~/.lm-studio -> ~/cache/lm-studio.

However, the custom -lm-dir flag did work for me! I installed gollama from brew today, version 1.30.4 so not sure if your fix is in already or not.

gollama -link-lmstudio
Scanning for LM Studio models in:
Failed to scan LM Studio models directory: LM Studio models directory does not exist:


% gollama -link-lmstudio -lm-dir ~/.cache/lm-studio/
Scanning for LM Studio models in: /Users/erg/.cache/lm-studio/
Found 7 LM Studio models
Processing model L3.2-Rogue-Creative-Instruct-Uncensored-Abliterated-7B-D_AU-Q4_k_s... success!
Processing model josiefied-qwen2.5-14b-instruct-abliterated-v4.Q4_0... success!
Processing model llama-3.2-3b-instruct-q8_0... success!
Processing model DeepSeek-R1-Distill-Qwen-7B-Q4_K_M... success!
Processing model Qwen2.5-14B-Instruct-Q4_K_M... success!
Processing model phi-4-Q4_K_M... success!
Processing model phi-4_Q5_K_M... success!

Summary: 7 models linked successfully, 0 failed


gollama -v
1.30.4