vndangkhoa/open-design
1
0
Fork 0
mirror of https://github.com/nexu-io/open-design.git synced 2026-05-31 19:04:39 +07:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

ci: retry PR-context gh calls so a transient API blip doesn't abort the run (#3128)

Browse source 
* ci: retry PR-context gh calls so a transient API blip doesn't abort the run

The early PR-context gathering calls `gh pr diff`, `gh pr view`, and
`gh api .../files`. gh hits api.github.com under the hood, and a single
transient timeout/5xx there aborts the whole run before any exploration
(seen on #3083: "could not find pull request diff: Get \"https://api.
github.com/...\": net/http timeout"). These were the only network calls
in the run without a retry (source fetch + npm already retry).

Add a small gh_retry helper (4 attempts, linear backoff) and wrap the
three read-only context calls. gh writes nothing to stdout on a failed
API call, so retrying is safe even for the calls piped into the context
file; the retry warning goes to stderr.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* ci: address review — buffer gh retries to files (no paginated duplication)

Review (Siri-Ray): wrapping `gh api --paginate` retries inline while the
context block is redirected to the file means a mid-pagination failure
leaves partial pages in the context, and the retry appends them again —
duplicating the patches section and burning the context budget the agent
reads.

Replace the in-pipe gh_retry with gh_retry_file: each call buffers to its
own file per attempt (`>` truncates on open, so a failed/partial attempt
is discarded before the next), and the context block just cats the
finished files. Fetch PR body + patches to files up front, then assemble.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
lefarcen 2026-05-27 22:42:08 +08:00 committed by GitHub
parent a6a56099ca
commit 54f225d6b3
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 27 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

31
.github/scripts/agent-pr-explore-sandbox.sh vendored
View file

@ -933,7 +933,23 @@ cat > "$artifacts/manifest.json" <<JSON
}
JSON
gh pr diff "$PR_NUMBER" --repo "$BASE_REPO" --name-only > "$changed_files_file"
# gh hits api.github.com under the hood; a single transient blip there
# (timeout / 5xx) would otherwise abort the whole run before exploration.
# Retry each read-only PR-context call, buffering its output to a file per
# attempt (> truncates on open) so a partial/paginated failure cannot
# duplicate output into the context the agent later reads.
gh_retry_file() {
local out="$1"; shift
local attempt
for attempt in 1 2 3 4; do
if "$@" > "$out"; then return 0; fi
[ "$attempt" = 4 ] && return 1
echo "::warning::gh call failed (attempt ${attempt}/4): $* — retrying" >&2
sleep $((attempt * 4))
done
}
gh_retry_file "$changed_files_file" gh pr diff "$PR_NUMBER" --repo "$BASE_REPO" --name-only
while IFS= read -r changed_path; do
if is_app_surface_path "$changed_path"; then
@ -951,6 +967,14 @@ echo "$agent_fixture" > "$artifacts/agent-fixture.txt"
deterministic_verifier="$(select_deterministic_verifier)"
echo "$deterministic_verifier" > "$artifacts/deterministic-verifier.txt"
# Fetch PR body + patches to files first (buffered retry), then assemble the
# context from the files so a retried/paginated call can never duplicate output.
pr_body_file="$artifacts/pr-body.txt"
gh_retry_file "$pr_body_file" gh pr view "$PR_NUMBER" --repo "$BASE_REPO" --json title,body --jq '"# " + .title + "\n\n" + (.body // "")'
pr_patches_file="$artifacts/pr-patches.txt"
gh_retry_file "$pr_patches_file" gh api --paginate "repos/${BASE_REPO}/pulls/${PR_NUMBER}/files" --jq \
'.[] | "### " + .filename + " (" + .status + ", +" + (.additions | tostring) + "/-" + (.deletions | tostring) + ")\n```diff\n" + (if .patch == null then "[binary or generated patch omitted]" else (.patch[0:'"$file_patch_max_chars"'] + (if (.patch | length) > '"$file_patch_max_chars"' then "\n[patch truncated]" else "" end)) end) + "\n```\n"'
{
echo "# PR #$PR_NUMBER context"
echo
@ -960,14 +984,13 @@ echo "$deterministic_verifier" > "$artifacts/deterministic-verifier.txt"
echo "Head SHA: $HEAD_SHA"
echo
echo "## PR body"
gh pr view "$PR_NUMBER" --repo "$BASE_REPO" --json title,body --jq '"# " + .title + "\n\n" + (.body // "")'
cat "$pr_body_file"
echo
echo "## Changed files"
cat "$changed_files_file"
echo
echo "## Text patches"
gh api --paginate "repos/${BASE_REPO}/pulls/${PR_NUMBER}/files" --jq \
'.[] | "### " + .filename + " (" + .status + ", +" + (.additions | tostring) + "/-" + (.deletions | tostring) + ")\n```diff\n" + (if .patch == null then "[binary or generated patch omitted]" else (.patch[0:'"$file_patch_max_chars"'] + (if (.patch | length) > '"$file_patch_max_chars"' then "\n[patch truncated]" else "" end)) end) + "\n```\n"'
cat "$pr_patches_file"
} > "$context_file"
head -c "$context_max_bytes" "$context_file" > "$trimmed_context_file"
if [ "$(wc -c < "$context_file" | tr -d " ")" -gt "$context_max_bytes" ]; then