Hey Bruno,
Bruno Haible via Gnulib discussion list <[email protected]> writes:
> 1) Code included in this package that comes from a single LLM prompt
> must be limited in size: it must be at most 5 lines long.
A user on Mastodon was under the impression that this policy means you
vibe code a large patch with multiple prompts, as long as each prompt
only produces 5 lines [1]. I.e., a 500 line patch is fine as long it was
the summation of 100 prompts creating diffs of 5 lines each.
Perhaps the "from a single LLM prompt" should be replaced with just
"from a LLM"? Since I don't think that was the intent of the policy. At
least, speaking personally, I would not accept a patch like that.
LLMs aren't really my thing, so it may be my misinterpretation, but I
feel like the "single prompt" may be unimportant/misleading. With
ChatGPT, you typically use the responses API which in their own words
[2]:
Create stateful interactions with the model, using the output of
previous responses as input.
Basically, it sounds like there is just a state on the server to avoid
long inefficient API calls like this which resend the entire
conversation each time, with a new message appended like this:
[
{
"role": "user",
"content": "...."
},
{
"role": "assistant",
"content": "..."
},
[...]
]
A patch will be created from the summation of this context, which is why
I feel that individual prompts may not be important.
Collin
[1] https://mastodon.social/@[email protected]/117209950774935489
[2] https://developers.openai.com/api/reference/responses/overview
