fix(buffers): assertion failed when buflist is nil

[phanen]

Problem: contents callback calls utils.CTX() without includeBuflist,
so if the context was recreated between core.lua:346 (which passes
includeBuflist=true) and the async contents invocation, buflist is
nil and the assertion crashes. The same file already handles this at
line 282 with utils.CTX().buflist or {}.

Solution: pass { includeBuflist = true } explicitly, ensuring the ctx
always has buflist populated when accessed inside the callback.

Closes #2736

Summary by CodeRabbit

Release Notes

No user-visible changes in this update. Internal implementation adjustments have been made to improve code reliability.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: fa190fb3-19e5-4419-a6f6-92211ba76bac

📥 Commits

Reviewing files that changed from the base of the PR and between 267f5db and fb07c7f.

📒 Files selected for processing (1)

• lua/fzf-lua/providers/buffers.lua

---

📝 Walkthrough

Walkthrough

In M.buffers, the contents closure changes how buflist is retrieved: instead of assert(utils.CTX().buflist), it now calls utils.CTX({ includeBuflist = true }).buflist, passing an option to include the buffer list in the context rather than asserting its presence.

Changes

Buflist context retrieval fix

Layer / File(s)	Summary
buflist retrieval in M.buffers lua/fzf-lua/providers/buffers.lua	buflist is now obtained via utils.CTX({ includeBuflist = true }).buflist instead of assert(utils.CTX().buflist), removing the hard assertion that caused intermittent failures when buflist was absent from the context.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

Poem

A bunny once stumbled on an assert so mean,
That crashed without warning mid-hop on the green.
Now CTX gets a nudge — "include the buflist, please!"
No more panics at dusk, no more failures to tease.
🐇✨ Smooth hops all around, with no error to seize!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: fixing an assertion failure when buflist is nil in the buffers provider.
Linked Issues check	✅ Passed	The code change directly addresses issue #2736 by preventing the nil assertion failure through explicitly passing includeBuflist=true parameter.
Out of Scope Changes check	✅ Passed	The change is narrowly scoped to fixing the specific assertion failure in buffers.lua by modifying how buflist is retrieved, with no extraneous modifications.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/buffers

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[ibhagwan]

So you found it :)

[phanen]

no, it's still a llm slop, I don't understand the real reason. Give llm more time to persuude me...

[ibhagwan]

GLM 5.2 is also convinced this is the right fix though I couldn’t reproduce this with its repro steps even when adding uv.sleep to make the refresh slower:

[phanen]

Fun fact I never reproduce since then..