# Large reference test datasets (git-LFS)

The `[large]` Catch tests in `tests/` run the full analysis/processing pipeline over real
JUNGFRAU datasets that are too big to keep as ordinary git blobs. They are tracked with
**git-LFS** (see the `tests/data/*.h5` rule in the top-level `.gitattributes`).

These files are **not required** to build or to run the normal test suite: every test that
needs them resolves the path through `jfjoch_test::LargeDataFile()` (`tests/TestData.h`) and
`SKIP()`s when the file is absent or is still an unfetched LFS pointer. `jfjoch_test` also
prints, at start-up, whether this directory is populated.

## Fetching

```
git lfs install
git lfs pull            # or: git lfs pull --include "tests/data/*.h5"
```

## Expected files

Place each dataset's master **and its `_data_NNNNNN.h5` files** in this directory:

| File                          | Dataset                                  |
|-------------------------------|------------------------------------------|
| `lyso_rotation_master.h5`     | lysozyme rotation series (~1800 images)  |
| `lyso_serial_master.h5`       | lysozyme serial / jet stills (~5000 images) |

To point the tests at an existing local dataset without copying, symlink it:

```
ln -s /path/to/your_master.h5            tests/data/lyso_rotation_master.h5
ln -s /path/to/your_data_000001.h5       tests/data/...
```

(The master references its data files by relative name, so keep them side by side.)