I'm encountering a consistent rendering issue with ResearchRabbit's visualizations (citation network, co-authorship maps) in Safari 17.5 on macOS Sonoma. The canvas area loads but remains entirely blank. The same workspace and visualization loads correctly in Chrome and Firefox on the same machine.
This is a significant blocker for my workflow, as Safari is my primary browser. The developer console in Safari shows the following error, which appears to be related to WebGL or canvas rendering:
```
[Error] WebGL: INVALID_OPERATION: texImage2D: incompatible format/type combination for this format
```
**Troubleshooting steps I have already performed:**
* Cleared Safari cache and website data for ResearchRabbit.
* Disabled all content blockers and Safari extensions.
* Confirmed `WebGL 2.0` is enabled in Safari's Develop menu.
* Tested on a different network to rule out CDN issues.
* The issue persists across multiple visualizations and collections.
The specific error points to a texture handling problem in the rendering pipeline. Given that Chrome works, this suggests a Safari-specific compatibility issue with the visualization library (likely Three.js or similar).
Has anyone else replicated this? A temporary workaround is to use another browser, but for a long-term research tool, browser compatibility is critical. I would be interested to know if the ResearchRabbit team is aware of this Safari compatibility matrix issue, and if there is a known fix or configuration change on the user side.
Right-size or die
Ah, that specific WebGL texture format error is a classic Safari pain point. It's almost certainly a bug in their WebGL implementation that the visualization library is tripping over. Chrome and Firefox are more permissive with certain format/type combos.
Your diagnosis about it being a Three.js or similar library issue is spot on. The devs likely used a texture format that Safari's renderer rejects as invalid, while other browsers silently accept it. I've seen this exact error pop up when a library tries to use `RGB` format with `UNSIGNED_BYTE` type on a texture that has internal format constraints, or when mipmaps are generated incorrectly.
The immediate workaround is to enable the WebGL 2.0 prototype in Safari's Develop > Experimental Features menu, if you haven't already. Sometimes the newer pipeline handles these textures differently. Failing that, you're looking at a code-level fix where the library needs to adjust its texture parameters for Safari's stricter compliance. You might want to file a bug with ResearchRabbit and include that exact console error - it's their cue to add a browser-specific condition for texture creation.
APIs are not magic.
Yep, that WebGL 2.0 prototype trick has saved me more than once. It's a bit of a band-aid, but it usually gets things moving.
Totally agree the bug report needs that exact console error. I'd also suggest they check if it's related to the new-ish OffscreenCanvas API. I've seen Safari throw similar texture errors when a library tries to use it for background processing without proper feature detection first.
Automate everything.