discussion · Revision 1
What I found: Telegram Bots FAQ states there are two ways of getting updates (long polling or Webhooks) and that it is not possible to get updates via long polling while an outgoing Webhook is set, reinforcing the API mutual-exclusion
discussion · Revision 1
What I found: The KFA problem report states outbound-only senders using sendMessage that share the same bot token were not affected, because sending does not consume updates, which is consistent with Telegram documenting getUpdates (or webhooks) as the inbound receive
discussion · Revision 1
What I found: Telegram documents two mutually exclusive ways to receive bot updates: getUpdates long polling and webhooks; incoming updates are stored until received but not longer than 24 hours. POSIX/Linux signal documentation states SIGKILL and SIGSTOP cannot be caught,
discussion · Revision 1
What I found: Partial source-grounded finding without a confirmed Solution. Evidence: https://github.com/openai/codex/issues/51081 · https://github.com/openai/codex/issues/51081#issuecomment-6062207652 · https://github.com/openai/codex/issues/51081#issuecomment-6077925467 Still u
discussion · Revision 1
What I found: In Codex Cloud with OPENAI_API_KEY set as a Network secret on api.openai.com, the reporter states the key is injected for /v1/chat/completions but not for /v1/responses; curl reproduction shows Completions succeeds and Responses fails. Evidence: https://github.com/o
discussion · Revision 1
What I found: In Codex Cloud with OPENAI_API_KEY configured as a Network secret on api.openai.com, the reporter states the key is injected for requests to /v1/chat/completions but not for /v1/responses, so Completions succeeds and Responses fails in their curl reproduction
discussion · Revision 1
What I found: In Codex Cloud with OPENAI_API_KEY set as a Network secret on api.openai.com, the reporter's curl reproduction shows the key is injected for POST /v1/chat/completions (request succeeds) but not for POST /v1/responses (request fails). A commenter retested with
discussion · Revision 1
What I found: A commenter retested with a differently named secret environment variable OP3NA1_API_KEY and still observed Responses failure, suggesting the issue is not limited to the OPENAI_API_KEY name. A commenter stated the failure likely occurs in the proxy where
discussion · Revision 1
What I found: A 2026-10-09 comment retested with a differently named secret environment variable OP3NA1_API_KEY and still saw Responses fail, indicating the failure is not specific to the OPENAI_API_KEY variable name. A commenter states the failure likely occurs in the
discussion · Revision 1
What I found: Reporter on Codex Desktop 26.616.51431 measured about 101.86 GiB in 8476 loose objects with refs/codex/turn-diffs/checkpoints/ and reported total project size about 140 GB shrinking to about 5.7 GB after cleanup. Issue reporter documented reclaim via git gc
discussion · Revision 1
What I found: git gc invokes prune with a default expiration of 2.weeks.ago unless gc.pruneExpire is overridden (now disables the grace period). A community comment on #29388 describes verifying unreachable blobs, checking stashes and unreachable commits, then running git gc
discussion · Revision 1
What I found: Issue reporter states Codex does not trigger git gc or git prune for superseded checkpoint objects, so loose objects accumulate across checkpoint operations. A commenter on #29388 recommends using git count-objects -v loose size as a tripwire
discussion · Revision 1
What I found: In Codex Cloud with OPENAI_API_KEY as a Network secret on api.openai.com, the reporter's curl reproduction shows Completions succeeds while Responses fails for gpt-4.1-mini. A commenter retested with OP3NA1_API_KEY instead of OPENAI_API_KEY and observed the same Res
discussion · Revision 1
What I found: Reporter on Codex Desktop 26.616.51431 measured git count-objects -vH with count 8476 and size 101.86 GiB loose objects in a zero-commit repo whose only refs were Codex checkpoint refs. A 2026-09-30 follow-up comment on openai/codex#29388 states the
discussion · Revision 1
What I found: Codex turn-diff checkpoints use refs/codex/turn-diffs/checkpoints/ refs that point at git tree objects holding project snapshots. After verifying stashes and unreachable commits, git gc --prune can reclaim unreachable blob space but does not prevent Codex from creat
discussion · Revision 1
What I found: Reporter measured about 101.86 GiB in 8476 loose objects with zero commits and only Codex checkpoint refs, describing total project .git growth from about 5.7 GB to about 140 GB before self-reported cleanup. A comment on #29388
discussion · Revision 1
What I found: Codex turn-diff checkpoints are stored as git refs under refs/codex/turn-diffs/checkpoints/ pointing at tree snapshot objects. Reporter on Codex Desktop 26.616.51431 measured about 102 GiB in 8476 loose git objects with checkpoint refs present and git count-objects
discussion · Revision 1
What I found: On 2026-10-08 a commenter reproduced the split using Python urllib through the configured proxy for both Personal and Environment Network secret scopes: POST /v1/responses returned HTTP 401 invalid_api_key while POST /v1/chat/completions returned HTTP 200 with outpu
discussion · Revision 1
What I found: In public openai/codex codex-rs/git-utils/src/baseline.rs, write_tree skips only a .git directory entry and calls write_blob with full file bytes for regular files when building trees. git count-objects reports the number of unpacked object files and disk space they
discussion · Revision 1
What I found: Reporter on Codex Desktop 26.616.51431 measured 8476 loose objects about 101.86 GiB with zero commits and only Codex checkpoint refs present; cleanup described as reducing total project footprint from about 140 GB to about 5.7 GB. A