karthink

Just a note to say that I haven't lost track of this, I'll get to it soon. I want to test it carefully as there are always unexpected edge cases...

> This creates a cyclic dependency: loading gptel autoloads the backend, which tries to load gptel. That can't quite be the problem, because by this logic the recursive load would...

> Is there a reason the gptel-openai backend has one dependency design and all the other backends have a different one? See #227 > allows customization of gptel-backend to succeed...

> If we autoload the gptel-backend struct A call to the autoloaded `gptel-make-ollama` (say) will make a call to the autoloaded `gptel-backend` which will load `gptel-openai` -- depending on a...

> Since gptel-make-... asks for a NAME, why not allow gptel-backend to simply refer to that name? So for instance (gptel-make-ollama "Ollama") and (setq gptel-backend 'Ollama) instead of (setq gptel-backend...

> [Also it occurred to me now that since gptel-make-* not only allocates a struct but also adds it to some kind of namespace, it would perhaps make sense to...

Just a note that if you want to avoid this issue while using `:custom`, you can do it by calling `gptel-make-*` in the use-package block of the corresponding feature. For...

> the docs indicate that youtube feeds are "org-linkable". Where do the docs imply this? > Since this is a playlist, I took the basic URL from the playist share...

Yeah, this is all over the tech Internet today. It appears to be a way to formalize tool-use for Anthropic models. I think I'm some distance away from adding support...

To be clear, I think the issues we're discussing in #481 and #475 are mostly about the UI and orthogonal to this feature. We would still have the same decisions...