C-x C-s in a notes buffer causes the position to be reset

[swflint]

With ebib-show-notes-method is set to all, upon save, the point, mark, narrowing, and tree-collapse state of the notes buffer is reset. It looks like this is being done by ebib--update-entry-buffer, but it's not clear why.

[joostkremers]

Could you elaborate a bit more? What steps do I need to reproduce the issue?

[swflint]

It looks like it may be a combination of the note file's buffer being killed (unclear what the reason for that is), and a call to goto-char in ebib-open-note. That being said, here is a minimum configuration to replicate this behavior:

(add-to-list 'load-path (locate-user-emacs-file "compat"))
(add-to-list 'load-path (locate-user-emacs-file "async"))
(add-to-list 'load-path (locate-user-emacs-file "parsebib"))
(add-to-list 'load-path (locate-user-emacs-file "citar"))
(add-to-list 'load-path (locate-user-emacs-file "helm"))
(add-to-list 'load-path (locate-user-emacs-file "ebib"))

(require 'ebib)
(setopt ebib-notes-new-note-hook nil
        ebib-notes-open-note-after-hook nil
    ebib-notes-show-note-method 'all
    ebib-notes-directory "~/literature-notes/")

Then, open a bibliography buffer for which you have at least one entry with notes in ~/literature-notes/, navigate to the notes buffer, make a change at the end of the buffer, and save with C-x C-s. Note that point is now at (point-min).

[joostkremers]

Hmm, yes, I see... What's happening is that when a note is opened, after-save-hook is set to ebib--maybe-update-entry-buffer, which calls ebib--update-entry-buffer, which in turn kills the note buffer when ebib-notes-storage is set to one-file-per-note (which is the default).

I'm not entirely sure why I did it that way, but I suspect it was just to avoid having dozens of buffers open with just a single note in them. If that's the case, then it would be an easy fix, but I'll have to dive into the code a bit more to be certain.

[swflint]

I figured it was something along those lines. Would it be possible to make that behavior configurable?

[joostkremers]

I've just pushed a commit that removes the kill-buffer from ebib--update-entry-buffer. I don't think it has any adverse effect except the one I mentioned. But please keep an eye out for any issues, because I don't use the one-file-per-note setting myself, so I can't really test it.

A user option is a good idea. If users complain that so many note buffer remain open, I'll implement it.

[swflint]

The buffer doesn't appear to be killed, but point is still being reset to (point-min). I think I saw somewhere in the note display code that it always resets the point location.

[swflint]

Wait, it seems like it's inconsistent. But I'm not sure if it isn't updating point.

[swflint]

Point is definitively still being reset.

[swflint]

Additionally, the entry buffer (not the notes buffer) still receives focus after save.

[joostkremers]

Point is definitively still being reset.

This should now be fixed with the latest update.

Additionally, the entry buffer (not the notes buffer) still receives focus after save.

This requires a more fundamental change. We'd need a way to update the contents of the entry buffer without killing and recreating the window displaying the note. Allow me a little bit of time to decide whether to just bolt that onto the existing function ebib--update-entry-buffer, or whether there's a better way to do it. (ebib--update-entry-buffer is already a bit of a hack, in my opinion...)

[joostkremers]

I added an optional argument to ebib--update-entry-buffer to keep the note window, which is now used in the after-save-hook in note buffers. This in itself doesn't change the focus issue, but before I work on that, I'd like to be sure that things still work the way they did before this change. Could you please check and let me know?

[swflint]

It seems to work! Thank you!

[joostkremers]

It seems to work!

Including the focus issue?

[swflint]

Yes!