Out variants of following ops are causing extra recompilations (in 3rd iteration) in torch.compile when compared to not using out variant, torch.bmm torch.topk torch.cholesky torch.linalg.norm torch.max torch.topk torch.bmm torch.cholesky torch.linalg.norm cc @chauhang @penguinwu @ezyang @bobrenjc93 @aditvenk @laithsa…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issuetriaged
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
when TORCH_DISTRIBUTED_DEBUG = detail, we should also enable cpp stacktraces so that if a collective is inconsistent, user can get the entire C++ stack to help debug. Arguably, we could also just do this for tdd = info. We could also discuss whether if TDD = detail, we also want to enable NCCL_DEBUG=warn or info, alth…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issueoncall: distributed
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
Right now we have this in tlparse: which often contains duplicated graph breaks: The goal is to have a summary of _distinct_ graph breaks for end users who just want to fix their graph breaks. cc @chauhang @penguinwu @voznesenskym @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisu…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issuetriaged
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
We should update the warning message "Can't remap async version of inplace op to functional collective" to point out that users can rewrite their own code to use the functional collective. (possibly point them to a doc with more helpful examples). Also, we might want to open a placeholder issue for adding dynamo suppo…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issueoncall: distributed
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
There are a lot of flake8-bugbear code reported by flake8-bugbear and ruff that need to be fixed. These would be a good starting issue for a new contributor. This can be found be removing the codes from the ignore list in the .flake8 file and the .pyproject file for ruff.
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issuemodule: lint
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
When using TORCH_DISTRIBUTED_DEBUG=DETAIL we collect collectives fingerprints and those are quite helpful when troubleshooting issues like stragglers. One recurring problem in distributed jobs are stragglers and, in special, those triggered by python GC activity. We should extend CollectiveFingerPrint to include two p…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issuetriaged
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
These enhancements are to have a better UX when using foreach_map, suggested in a few places, but most recently, https://github.com/pytorch/pytorch/issues/158371#issuecomment-3088757068 These enhancements should allow easier compiler-first custom optimizer implementations. cc @chauhang @penguinwu @voznesenskym @EikanW…
标题与摘要为来源原文（英文），暂无官方中文版本。
pytorch/pytorch约 4 小时good first issuefeature
已交叉核验 · GitHub Good First Issues · ai核验于 2026年8月25日加入路径
Do you want to request a feature or report a bug? Bug What is the current behavior? I'm seeing "Error: Should not already be working" after upgrading to React 16.11 Which versions of React, and which browser / OS are affected by this issue? Did this work in previous versions of React? This is exclusively happening on…
标题与摘要为来源原文（英文），暂无官方中文版本。
react/react约 4 小时good first issueType: Bug
已交叉核验 · GitHub Good First Issues · frontend核验于 2026年8月25日加入路径
Hello, Currently, you can define a new connection/variable in several ways. Metabase Environment variable Secret backend However, only the connections defined with the Metabase are visible from the Web UI. This is surprising among users because a natural way to check connection/variable configuration is to use the Web…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
The SFTPOperator currently runs in synchronous mode only, blocking a worker slot for the entire duration of every file transfer. For large file transfers, this wastes worker resources unnecessarily. Proposed Solution Add deferrable=True parameter to SFTPOperator following the same pattern used in other Airflow provide…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuearea:providers
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
depends_on_past=True currently allows a task to depend only on the same task in the previous dag_run. There is no native way for a task to depend on the successful completion of multiple specific tasks from the previous dag_run (within the same DAG). Achieving this today requires custom sensors that directly query Air…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
We do not have triggers information exposed in provider info but some Providerst already expose some (for example cncf.kubernetes). What needs to be done:
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:meta
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
Enable breeze to utilize podman for container management. Given the open source nature, lack of licensing issues, OCI-compliant images, and improved security and performance, enable podman as a container management option for breeze tooling.
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
GlueJobOperator._find_job_run_id_by_task_uuid (the task-UUID scan fallback, used when neither task_state_store nor a cached XCom id has a prior run) has a few pre-existing quality issues, carried over verbatim from main, that are worth cleaning up: 1. Unbounded pagination. It's a while True over get_job_runs(MaxResult…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issueprovider:amazon
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
The default Airflow task failure email is difficult to read and not optimized for operators. Currently the subject and body contain a raw Python object dump (UUIDs, repr(), internal fields), which makes it hard to quickly understand: - which DAG failed - which task failed - when it failed - where to click for logs Thi…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuearea:providers
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
2.7.1 When defining an SMTP connection only the user and password are taken from the connection. The host, port, starttls, start_ssl, from_email, timeout and retry are all taken from the configuration and not from the connection. Although the smtp provider (https://airflow.apache.org/docs/apache-airflow-providers-smtp…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:bug
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
3.0.6 Running DatabricksRunNowOperator(deferrable=True) (and this likely applies to DatabricksSubmitRunOperator too, since both share the same trigger/hook code) submits the Databricks job successfully and the task defers correctly (worker slot released). But the triggerer then crashes while polling job status, before…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:bug
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
Currently in the ExternalTaskSensor there are two ways two set the execution date of a DAG/task the sensor is waiting on: - Provide the execution_delta which simply adds/subtracts a time value to/from the execution_date of the sensor itself - Provide the execution_date_fn which is a function that takes in just the exe…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
Search – Assets - Filter by Asset Group ↪ Narrow results to a logical grouping of assets (e.g., revenue_reporting, customer_metrics) - Filter by Consuming DAG and / or Producing Task ↪ Quickly locate assets by their upstream producer or downstream consumer - Filter by Last Event Timestamp (Range) ↪ Identify stale or r…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
The InfluxDB3Operator added in #58929 currently executes queries synchronously, which means each task holds an Airflow worker slot for the entire duration of the query. For long-running analytics or aggregation queries over InfluxDB 3 clusters, this prevents the worker from picking up other tasks while the query runs…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issuekind:feature
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径
related: #49938 Several users have pointed out that the Event-Driven Scheduling feature introduced in Airflow 3 currently supports only Amazon SQS. So I add Kafka common message queue provider in #49938, but there are still quite a few common providers that should be useful for airflow users to implement Event-Driven…
标题与摘要为来源原文（英文），暂无官方中文版本。
apache/airflow约 4 小时good first issueprovider:google
已交叉核验 · GitHub Good First Issues · data核验于 2026年8月25日加入路径