Daniel Mendler

@mhcerri Org-ql does essentially this: ~~~ elisp (let ((completion-styles '(basic))) (completing-read "Dynamic: " (completion-table-dynamic (lambda (str) ;; some dynamic candidates... (list (concat str "1") (concat str "2") (concat str "3"))))))...

@mhcerri what happens if you enter something?

@mhcerri No. My example doesn't work as expected with Ivy. I just tested this myself. For example if you enter "foo" the expected output should be: ~~~ foo1 foo2 foo3...

@mhcerri I see. The problem is that Ivy doesn't use completion styles for filtering, such that the `completion-styles` override here isn't effective. Hopefully it is sufficient if @alphapapa let-binds `ivy-completing-read-dynamic-collection`...

@SterlingHooten The nil error you've found is a bug in the org-ql completion table. If that is fixed org-ql should work perfectly well with default completion, vertico and icomplete.

Does this mean that pdf-tools is going to switch to a python backend in the long term, because the stated goal of this PR is to lower the maintainance barrier?...

This is just my personal opinion and I am not involved with this project, except that I am a happy user. But I see a problem if multiple backends with...

I understand your motivation and I agree that it would help if you want more people to look at your python backend. But let's be clear: This will not reduce...

I should remark that moving out the server only does not address the problematic consequences I mentioned before. I don't consider the mere presence of python code a problem, I...

> It is comparable to having different language servers available for LSP; the LSP client is implemented once in Emacs, then all servers that would like to use it should...