Panic when using openai

[4ydx]

I thought to try to submit a PR, but you have a git hook with cargo bin, which I'm not familiar with. Anyway, here is a - probably incorrect - first attempt:

diff --git a/src/domain/services/bubble.rs b/src/domain/services/bubble.rs
index 9af582b..bc5f3e9 100644
--- a/src/domain/services/bubble.rs
+++ b/src/domain/services/bubble.rs
@@ -218,7 +218,7 @@ impl<'a> Bubble<'_> {
                 return line.len();
             })
             .max()
-            .unwrap();
+            .unwrap_or(0);

         if max_line_length > (self.window_max_width - line_border_width) {
             max_line_length = self.window_max_width - line_border_width;

After the change above is applied and compiled, I am seeing that "loading..." never goes away after chat gpt-4 completes sending data. So there is something incompatible in the stream logic perhaps?

[dustinblackman]

Hey there! Thanks the report.

This sounds more like a bug in reading in the OpenAI responses rather than how bubbles are rendered. I'll take a look!

[4ydx]

Agreed. I simply wanted to stop the panic so that I could even begin to think about what might actually be going on.

Is it correct to assume that here you expect the loop to exit once the request is fully processed so that done will be marked as true? Am I reading that correctly?

[jim-at-jibba]

I assume I am getting the same error when using openai. Ollama models all work fine

[dustinblackman]

Sorry for the delay, fixed in v0.12.4. Thanks!