Daniel Hollas

> I am still not a 100% clear about the hypothetical case of the plugin being updated needing the cache to be invalidated (by the hash being different). Wouldn't the...

> Still, in this situation, one could "invalidate" the original CALCJOB_A node so it won't be used for the cache anymore. How common are these scenarios? Would it not be...

> If user try to enable the caching, we pop up a warning with links sort of like that "SHOULD READ BEFORE USING CACHING". I like that idea, we can...

Would be great to have this, I've just started using caching and it is perfect for our use case. > How common are these scenarios? Would it not be more...

> I did suggest to go on with removing all versions from the hash, but then @danielhollas voiced concerns. I responded to his suggestions, explaining why I think they are...

Note that we're currently using the `ComputerBuilder` from this module in AiiDAlab. https://github.com/aiidalab/aiidalab-widgets-base/blob/96d589d41a4edf7bc8e84f6048a06345b9d54455/aiidalab_widgets_base/computational_resources.py#L939 cc @unkcpz @yakutovicha

Hi @xiasummer, this plugin is now obsolete, and there is a new official version: https://meta.discourse.org/t/discourse-math-plugin/65770 The new plugin also support KaTeX renderer instead of MathJax, which is much faster and...

If it turns out to be too difficult to fix quickly, we could theoretically postpone this migration. Is it really needed to drop the hashes? All that will happen is...

Hi :wave: Just wanted to signal our interest in this PR. :blush: I've tested this branch and by my (very rough) benchmark, this would save us around 300 ms startup...

> To add to that - editable mode is tricky and I’d prefer / expect uv to rebuild it every time. Keep the current logic for non-editable packages such that...