[BUG] Panic when using `job-label: true` in query-comment config (BigQuery)

[dataders]

Describe the bug

Filing on behalf of a customer (Slack thread).

When using job-label: true in the query-comment config for BigQuery, dbtf compile panics at runtime. The panic occurs because the resolved query comment JSON may contain non-string values (e.g., nested objects like relation), and the code at query_comment.rs:128 calls value.as_str().expect("Expected job label to be a string") which panics on non-string JSON values.

There are two related issues:

1. Panic on valid config: job-label: true is a valid boolean value, but causes a panic when the resolved query comment contains non-string JSON values (objects, arrays, numbers, booleans).
2. Misleading validation feedback: Setting job-label: "some_string" correctly rejects with dbt1005: Invalid query comment: job-label must be a bool — but since the valid bool value true panics, this creates a confusing experience where neither bool nor string works.

What version of dbt Fusion is this bug in? (find out by running dbt --version)

dbt-fusion 2.0.0-preview.114

Is this a discrepancy between the dbt Fusion Engine and dbt Core? Check one.

• YES
• NO

To Reproduce

dbt_project.yml:

query-comment:
    comment: |
        {% if target.name == 'dev' %}
          {{ query_comment(node, 'dev') }}
        {% else %}
          {{ query_comment(node, env_var('DBT_EXECUTED_BY', 'other')) }}
        {% endif %}
    job-label: true

Run:

dbtf compile

Output:

dbt-fusion 2.0.0-preview.114
   Loading profiles.yml
  Fetching packages.yml
 Rendering [━━━━━━━━━━━━━━━━━━━━] 22697/22698 22697 succeeded | 1 in-progress
   unit_test:base_contract_renewal_dates [0.5s]
panic: panicked at fs/sa/crates/dbt-adapter/src/query_comment.rs:128:55:
Expected job label to be a string

Root Cause

In query_comment.rs:128, the get_job_labels_from_query_comment method iterates over the parsed JSON map and calls:

sanitize_label(value.as_str().expect("Expected job label to be a string"))

This assumes every value in the query comment JSON is a string, but query comments commonly contain nested objects (e.g., relation: {"database": "...", "schema": "...", "identifier": "..."}), numbers, or booleans. The .expect() call causes a panic instead of handling these gracefully.

The validation in resolve_query_comment.rs:41-47 correctly validates that job-label itself is a bool, but nothing validates or handles non-string values within the resolved query comment JSON when converting to BigQuery job labels.

Expected behavior
Non-string JSON values in the resolved query comment should be handled gracefully when job-label: true — either by converting them to strings (e.g., via to_string() / serde_json::to_string()) or by skipping/flattening nested values. This is how dbt Core + dbt-bigquery handles it via connections.py#L629.

Operating System and CPU Type (please complete the following information):

• Mac/Windows/Linux? macOS
• X86 or ARM? ARM

[VersusFacit]

seems to be addressed by https://github.com/dbt-labs/fs/pull/8002

[fornwall]

seems to be addressed by https://github.com/dbt-labs/fs/pull/8002

@VersusFacit Great! As that repository is private, do you know if that fix has been released (or when it will be)?