`request` should be able to retry upon broken connections
Metadata
Current evaluation
remote-build crashes with unhandled ChunkedEncodingError when artifact download breaks mid-stream; request service has no retry support. Still open, no maintainer response, code path unchanged.
Suggested action: needs triage
Reason: The issue has zero comments, no labels, no assignee, and no maintainer response in 249 days, so it has never been triaged despite being a well-written bug report with a full traceback. The defect is still present at HEAD: request.py line 71 streams via `download.iter_content(None)` with no retry and the Session has no HTTPAdapter/max_retries configured. However, the reporter proposes only the desired behaviour, not a concrete design, so implementing it needs maintainer decisions on retry scope and how to handle a partially written destination file and already-advanced progress bar.
Evaluated by: mr-cal-bot
Issue body
Evaluation history
| Date | Model | Scores | Action | Summary |
|---|---|---|---|---|
| stealth/space-bunny-alpha |
Impact:
52
Quick Win:
16
Complexity:
45
Confidence:
82
Actionability:
55
|
needs triage | remote-build crashes with unhandled ChunkedEncodingError when artifact download breaks mid-stream; request service has no retry support. Still open, no maintainer response, code path unchanged. | |
| stealth/space-bunny-alpha |
Impact:
55
Quick Win:
19
Complexity:
55
Confidence:
85
Actionability:
75
|
needs triage | RequestService.download_chunks has no retry, so a ChunkedEncodingError mid-download aborts remote-build artifact fetching and wastes a completed build. Untouched since filing: no labels, no comments, no maintainer response. | |
| stealth/space-bunny-alpha |
Impact:
55
Quick Win:
23
Complexity:
35
Confidence:
72
Actionability:
65
|
needs triage | RequestService.download_chunks has no retry, so ChunkedEncodingError from network flakiness fails the whole remote-build artifact/log fetch, discarding a successful build. Awaiting maintainer triage. | |
| qwen/qwen3.6-35b-a3b |
Impact:
65
Quick Win:
27
Complexity:
35
Confidence:
75
Actionability:
65
|
needs triage | Feature request to add retry logic for ChunkedEncodingError during artifact downloads in remote-build service | |
| qwen/qwen3.8-27b |
Impact:
55
Quick Win:
30
Staleness:
65
Complexity:
45
Confidence:
80
Support Request:
10
|
needs triage | Request service download_chunks crashes with unhandled ChunkedEncodingError when a remote-build artifact download breaks mid-transfer, forcing a full rebuild. No retry logic exists in the code at HEAD; issue is unlabeled with no maintainer activity for ~7 months. | |
| qwen/qwen3.6-35b-a3b |
Staleness:
70
Complexity:
45
Confidence:
90
Support Request:
10
|
needs triage | Feature request to implement retry logic for broken network connections during artifact downloads. Issue is unlabelled, has zero maintainer comments, and has been inactive for over 6 months. | |
| qwen3.6-35b-a3b-mtp-q6 |
Staleness:
75
Complexity:
40
Confidence:
90
Support Request:
10
|
needs triage | Request to implement retry logic for transient network errors during artifact downloads in craft_application. The issue is unlabelled, has zero comments, and has been inactive for 164 days. |
Update history
No update history recorded yet.
Related work
-
Related To:
craft-application#192
(confidence 70%)
Commit 4e2f7142 'feat: request service (#192)' introduced the download_chunks path that lacks retries.
-
Related To:
craft-application#278
(confidence 55%)
Prior art in the same repo for retrying flaky network calls (ac19c13e, b589a107, b97d76d6), which established the craft_application.util.retry helper this fix would likely reuse.
Related issues
No related issues found above the similarity threshold.