Search before asking
Paimon version
1.4
Compute Engine
python sdk
Minimal reproduce step
- With
FileSystemCatalog:
fs_catalog.alter_table(
Identifier("db", "table", branch="dev"),
[SchemaChange.add_column("branch_col", AtomicType("STRING"))],
)
Reload the main and dev tables. The new column is incorrectly added to main instead of only to dev.
- Create a table and a branch named dev. With RESTCatalog, load the branch table:
table = catalog.get_table(Identifier("db", "table", branch="dev"))
assert table.current_branch() == "dev"
assert table.schema_manager.branch == "dev"
The second assertion fails because schema_manager.branch is main.
What doesn't meet your expectations?
Schema operations should use the branch specified by Identifier. Altering a filesystem branch should not modify the main branch and a REST branch table should read schemas from that branch. Currently, both operations can incorrectly use the main branch.
Anything else?
No response
Are you willing to submit a PR?
Search before asking
Paimon version
1.4
Compute Engine
python sdk
Minimal reproduce step
FileSystemCatalog:Reload the main and dev tables. The new column is incorrectly added to main instead of only to dev.
The second assertion fails because schema_manager.branch is main.
What doesn't meet your expectations?
Schema operations should use the branch specified by Identifier. Altering a filesystem branch should not modify the main branch and a REST branch table should read schemas from that branch. Currently, both operations can incorrectly use the main branch.
Anything else?
No response
Are you willing to submit a PR?