KhronosGroup / KhronosGroup/OpenCL-Docs
A few clarification requests around image creation from another image
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 421
- Forks
- 132
- Avg merge
- 5d 13h
- Merged PRs (30d)
- 11
Description
- It seems the specification doesn't forbid creating an image from an image created from an image. This means implementations need to support an arbitrarily deep stack of image from image creations. Any objections to forbidding this?
- The table of supported image channel orders contains pairs of format associations except for
CL_DEPTHandCL_Rwhere the creation in only allowed in one direction. Is this an oversight or is there a good reason? - Nothing is said about creating an image from an image with the same channel order. This is arguably implicitly forbidden since there are no entries in the table of supported image channel orders. I'm not sure this is particularly useful but should it be allowed or forbidden? Some implementations have been accepting this.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the OpenCL specification's rules for creating an image from another image and the table of supported image channel orders. Investigate the three questions about recursive creation, the CL_DEPTH/CL_R direction, and matching channel orders. Done means the specification's behavior is explicitly resolved and its wording is updated accordingly.
Written by the indexing model from the issue text.
Assessment
- Domain
- documentation
- Issue type
- Documentation
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100