This walkthrough takes one through steps it takes to add a simple test to the library.
test-images directory is the repo that contains sample images that help test for regressions and correctness.
It contains a set of small/medium-sized images that stress certain parts of the decoder, so in case we have a
regression we don't want to occur, we add the image to test-images.
The structure is test-images->format->[images] for ease of use.
- Remember to add the images to VCS
E.g there is a jpeg.json which contains information about jpeg files.
The format is simple
| Field | Status | Explanation |
|---|---|---|
| name | Mandatory | File name with extension |
| hash | Mandatory | xxhash(seed of 0) result of the decoded result still not in planar format |
| colorspace | Optional | Expected colorspace output of the image, only used by zune-jpeg currently |
| comment | Optional | A reason why the file exists, useful for debugging purposes |
Here is an example of such a file description
{
"name": "weid_sampling_factors.jpg",
"hash": 290923794124024565485763288568005295883,
"comment": "Non normal sampling factors, h2v1 for all components",
"colorspace": "rgb"
}Each entry is part of an array, so it should be inside []
Tip: Incase you don't know the hash and you have ensured the output is correct(matches other libraries)
- Run the testing once, once you get the error, copy the hash printed to stdout
Ensure nothing is failing.
Now we can easily track regressions