gweaverbiodev

Using component without formally installing via pip

2

Is it possible to incorporate a custom component without wrapping it into a formal Python package? For context, we use [Bazel](https://docs.bazel.build/versions/main/be/python.html) for handling build dependencies and do not have formal...

enable setting default number of partitions to use for joins in `DaftContext`

3

**Is your feature request related to a problem? Please describe.** `shuffle_aggregation_default_partitions` already exists to set the number of partitions to some sane default for an entire job. It would be...

[Feature]: Support S3 endpoint with no prefix (http/https)

1

### Description Hi, We host our own internal S3 service in k8s that mimics the S3 interface. To properly route to this service, we need to provide the endpoint without...

jobs fail when setting concurrency on stateful UDFs on Ray runner

2

### Describe the bug When I set `concurrency` to some non-null value and run job on Ray, tasks are failing with logs like this one. ``` (pid=8503, ip=10.216.101.117) PhysicalScan->Project->DistributedActorPoolProject->DistributedActorPoolProject->DistributedActorPoolProject->DistributedActorPoolProject->CatalogWrite 0:...

write_iceberg should handle write.data.path and write.metadata.path properties if set

1

### Describe the bug currently, the assumption is that the [data path for iceberg tables is at `{location}/data`](https://github.com/Eventual-Inc/Daft/blob/ff936130558a21b6d61b9bc8165945f125c53b80/daft/logical/builder.py#L358), but this is not true if `write.data.path` is set and will break...