diff --git a/src/pypi_data/cli.py b/src/pypi_data/cli.py
index 86b277a1..ce500405 100644
--- a/src/pypi_data/cli.py
+++ b/src/pypi_data/cli.py
@@ -157,51 +157,6 @@ def merge_datasets(repo_path: Path, output: Path):
     asyncio.run(combine_parquet(repos, output))
 
 
-#
-#
-# @app.command()
-# def output_sql(repo_path: Path, redirects: Annotated[bool, typer.Option()] = False,
-#                batches_of: Annotated[Optional[int], typer.Option()] = None,
-#                settings: Annotated[str, typer.Option()] = '',
-#                dialect: Annotated[str, typer.Option()] = 'duckdb',
-#                insert: Annotated[bool, typer.Option()] = False):
-#     # log.info(f'Reading repos from {repo_path}')
-#     contents = _read_path(repo_path)
-#     repos = Repos.model_validate_json(contents)
-#
-#     if redirects:
-#         urls = asyncio.run(resolve_dataset_redirects(repos.root))
-#     else:
-#         urls = [r.dataset_url for r in repos.root]
-#
-#     if dialect == 'duckdb':
-#         def make_url(url: str):
-#             return f"read_parquet('{url}')"
-#     elif dialect == 'clickhouse':
-#         def make_url(url: str):
-#             return f"url('{url}', 'Parquet')"
-#     else:
-#         raise ValueError(f"Unknown dialect: {dialect}")
-#
-#     sql_queries = [
-#         f"SELECT * FROM {make_url(url)}"
-#         for url in urls
-#     ]
-#
-#     if batches_of:
-#         batches = list(itertools.batched(sql_queries, batches_of))
-#     else:
-#         batches = [sql_queries]
-#
-#     for batch in batches:
-#         query = 'INSERT INTO output\n' if insert else ''
-#         query += "\nUNION ALL ".join(batch)
-#         if settings:
-#             query += f'\nSETTINGS {settings}'
-#
-#         print(query + ';\n')
-
-
 async def resolve_dataset_redirects(
     repositories: list[CodeRepository], concurrency: int = 10
 ) -> list[str]:
diff --git a/src/pypi_data/combine_parquet.py b/src/pypi_data/combine_parquet.py
index 9c43647a..3df228ae 100644
--- a/src/pypi_data/combine_parquet.py
+++ b/src/pypi_data/combine_parquet.py
@@ -62,8 +62,6 @@ async def fill_buffer(
             log.info(f"Failed to download {repo.dataset_url}")
             continue
 
-        table = table.combine_chunks()
-
         for idx, batch in enumerate(table.to_batches(max_chunksize=2_500_000)):
             batch: RecordBatch
             digest = hashlib.sha256()
diff --git a/src/pypi_data/datasets.py b/src/pypi_data/datasets.py
index ef0cd0c4..b99c1971 100644
--- a/src/pypi_data/datasets.py
+++ b/src/pypi_data/datasets.py
@@ -35,7 +35,6 @@ class CodeRepository(pydantic.BaseModel):
     name: str
     url: HttpUrl
     ssh_url: str
-    parquet_url: HttpUrl
     index: RepositoryIndex | None = None
 
     @pydantic.computed_field()
@@ -62,7 +61,6 @@ def fetch_all(cls, github: Github) -> list[Self]:
                 name=repo.name,
                 url=repo.html_url,
                 ssh_url=repo.ssh_url,
-                parquet_url=f"{repo.html_url}/releases/download/latest/dataset.parquet",
             )
             for repo in github.get_organization("pypi-data").get_repos()
             if repo.name.startswith("pypi-mirror-")
@@ -127,16 +125,9 @@ async def download_dataset(self, client: httpx.AsyncClient) -> pa.Table | None:
         log.info(
             f"Loading parquet file with size {len(response.content) / 1024 / 1024:.1f} MB"
         )
-        return pq.read_table(pa.py_buffer(memoryview(response.content)))
-
-        # async with client.stream("GET", str(self.dataset_url)) as response:
-        #     if self.check_response(response, follow_redirects=True) is None:
-        #         return None
-        #
-        #     with output.open("wb") as f:
-        #         async for buffer in response.aiter_bytes():
-        #             f.write(buffer)
-        #     return True
+        return pq.read_table(
+            pa.py_buffer(memoryview(response.content))
+        ).combine_chunks()
 
     def without_index(self) -> Self:
         return self.model_copy(update={"index": None})