Copy does not create a new chunk

[ggerganov]

When a block is copied with Ctrl+C or Cmd+C, the copied text should be added as a chunk in the extra context. Currently, this does not seem to happen. To reproduce, I copy a chunk of text and way to see a request to the server, but there is no such request.

Version: 0.0.6

[igardev]

The chunk is added in the extra context and is visible if you get it with Ctrl+Shift+, . However, seems that there is a problem with the request. I will research it.

[igardev]

I've noticed that sometimes llama.cpp server enters a specific state and become unresponsive (not yet sure when and how). While in this state the log is:
srv update_slots: all slots are idle
request: POST /infill 172.22.224.1 200
slot launch_slot_: id 0 | task 343 | processing task
slot update_slots: id 0 | task 343 | new prompt, n_ctx_slot = 32768, n_keep = 0, n_prompt_tokens = 32186
slot update_slots: id 0 | task 343 | kv cache rm [0, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 1024, n_tokens = 1024, progress = 0.031815
slot update_slots: id 0 | task 343 | kv cache rm [1024, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 2048, n_tokens = 1024, progress = 0.063630
slot update_slots: id 0 | task 343 | kv cache rm [2048, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 3072, n_tokens = 1024, progress = 0.095445
slot update_slots: id 0 | task 343 | kv cache rm [3072, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 4096, n_tokens = 1024, progress = 0.127260
slot update_slots: id 0 | task 343 | kv cache rm [4096, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 5120, n_tokens = 1024, progress = 0.159075
slot update_slots: id 0 | task 343 | kv cache rm [5120, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 6144, n_tokens = 1024, progress = 0.190890
slot update_slots: id 0 | task 343 | kv cache rm [6144, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 7168, n_tokens = 1024, progress = 0.222706
slot update_slots: id 0 | task 343 | kv cache rm [7168, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 8192, n_tokens = 1024, progress = 0.254521
slot update_slots: id 0 | task 343 | kv cache rm [8192, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 9216, n_tokens = 1024, progress = 0.286336
slot update_slots: id 0 | task 343 | kv cache rm [9216, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 10240, n_tokens = 1024, progress = 0.318151
slot update_slots: id 0 | task 343 | kv cache rm [10240, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 11264, n_tokens = 1024, progress = 0.349966
slot update_slots: id 0 | task 343 | kv cache rm [11264, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 12288, n_tokens = 1024, progress = 0.381781
slot update_slots: id 0 | task 343 | kv cache rm [12288, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 13312, n_tokens = 1024, progress = 0.413596
slot update_slots: id 0 | task 343 | kv cache rm [13312, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 14336, n_tokens = 1024, progress = 0.445411
slot update_slots: id 0 | task 343 | kv cache rm [14336, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 15360, n_tokens = 1024, progress = 0.477226
slot update_slots: id 0 | task 343 | kv cache rm [15360, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 16384, n_tokens = 1024, progress = 0.509041
slot update_slots: id 0 | task 343 | kv cache rm [16384, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 17408, n_tokens = 1024, progress = 0.540856
slot update_slots: id 0 | task 343 | kv cache rm [17408, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 18432, n_tokens = 1024, progress = 0.572671
slot update_slots: id 0 | task 343 | kv cache rm [18432, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 19456, n_tokens = 1024, progress = 0.604486
slot update_slots: id 0 | task 343 | kv cache rm [19456, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 20480, n_tokens = 1024, progress = 0.636302
slot update_slots: id 0 | task 343 | kv cache rm [20480, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 21504, n_tokens = 1024, progress = 0.668117
slot update_slots: id 0 | task 343 | kv cache rm [21504, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 22528, n_tokens = 1024, progress = 0.699932
slot update_slots: id 0 | task 343 | kv cache rm [22528, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 23552, n_tokens = 1024, progress = 0.731747
slot update_slots: id 0 | task 343 | kv cache rm [23552, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 24576, n_tokens = 1024, progress = 0.763562
slot update_slots: id 0 | task 343 | kv cache rm [24576, end)
slot update_slots: id 0 | task 343 | prompt processing progress, n_past = 25600, n_tokens = 1024, progress = 0.795377
^C

There are other cases when the chunk is not added and there is no request - if it is too small or if it is identical with an existing chunk, but most probably this is not the reason for this bug.

[ggerganov]

@igardev This is normal behavior when there are more than one VS Code instances running at the same time. Each one would have a different context accumulated in it's ring buffer and when you switch between them, the llama-server would need to recompute the entire context instead of reusing it. This can become very slow for contexts of several thousand tokens.

[igardev]

@ggerganov Yes, I have two running instances of VS Code. We could add a remark about that in the ReadMe.md file.