Nexus
Images or file resources cannot be read
Check publication, expiry, and whether a disk path was mistaken for a web URL.
Identify the resource category
| Observation | Diagnosis and response |
|---|---|
| Images or scripts packaged with the page | Check relative paths and package manifests first. See Page-resource troubleshooting. |
| Python-published images, files, or meshes | Obtain a valid reference for the current page, then read its content. |
| Files the user just picked or dropped | Inspect per-item import success and rejection reasons. Receiving a reference does not prove decoding or saving succeeded. |
References have scope and lifetime
Real disk paths, file:// URLs, and another add-on's references are not universal page addresses. Resources belong to their publisher's add-on, client, and document scope. Reloading or closing can invalidate references. Use the actual requesting window's context when publishing to multiple windows.
Distinguish references from temporary reading URLs. Renew an expired URL from a still-valid reference; republish or reimport if the reference expired. Do not edit timestamps or retry indefinitely.
Match decoding to the data
| Observation | Diagnosis and response |
|---|---|
| Ordinary image fails to display | Check MIME, actual file headers, publication method, and decoding errors. Do not just change extensions. |
| Floating-point image cannot open as a normal picture | rgba32f contains floating-point pixels, not PNG data. Use the matching reader rather than assigning it directly to img. |
| Mesh returned but model is invisible | A mesh resource is data, not an object automatically added to the web scene. Construct and render it from metadata. |
| Upload did not change Blender | Uploading only stages content. Writes also require a registered action, permissions, valid targets, and result handling. |
For supported sources, formats, and examples, see Resources and file configuration. Do not use ordinary file/bytes to bypass image publication constraints.
Picking a file is not import or save completion
User cancellation is normal, not a network failure. Each import has its own lifetime. Retained page data differs from unread references; do not persist temporary references in project configuration.
Type, count, and byte budgets affect imports; show per-item rejection reasons. Save destinations are scoped references, not paths. Obtaining one does not complete saving. Check submission results and overwrite policies. See File import and File export.
Verify reads and application results
- Inspect actual content or images, not only the existence of a resource ID.
- Display each success, rejection, or cancellation in the workflow.
- Obtain fresh resources after reopening and handle expired addresses correctly.
Do not attach an entire private project. Prefer small, nonsensitive reproductions.
If unresolved, follow Provide useful diagnostic information to prepare a minimal reproduction and original messages for the add-on author.