bufler-workspace-tabs-mode breaks tab-bar--current-tab-index about bufler.el HOT 4 CLOSED

Thanks for the kind words. Please keep the feedback coming! I want to tag 0.2 and start making some other improvements, but I want to get these bugs that only come out with wider testing ironed out first. Couldn't do it without you. :)

from bufler.el.

Hi Nick (or should I say No Wayman?),

Another great package! bufler-workspace-tabs-mode looks particularly promising.

Thanks.

I've been experimenting with it and ran into a minor issue:
bufler-workspace-tabs seems to break tab-bar--current-tab-index

The backtrace shows that the frame variable is nil instead of being the value of (selected-frame). I reproduced this on Emacs 28 by evaluating (tab-bar--current-tab-index), but I can't explain why the frame variable is nil instead of the value of (selected-frame) as defined in the cl-defun form. Even the backtrace shows the code that should set the variable.

I haven't had much time to look into the source, but I did note your commentary on bufler-workspace-tabs and the complicated requirements of its implementation.

The complexity is entirely internal to Bufler and the way I implemented group paths. The issue is that paths can begin with and include nil elements, but those nil elements must be omitted when displaying paths, yet they must be preserved for accessing buffers by path. When I made that feature, I didn't realize how many places would need to account for it, and how much whack-a-mole I'd have to play when working on other features.

What initially brought this to my attention was doom-modeline.
It has a segment that displays the current tab index in the modeline and it signals (quite a few depending on the situation) errors on redisplay:

Error during redisplay: (eval (doom-modeline-segment--workspace-name)) signaled (wrong-type-argument framep nil) [1038 times]

If there's anything more I can do to help, let me know.

If you could figure out why the cl-defun-defined function is not setting the frame variable, that would be great. I have no idea what's going on there.

from bufler.el.

Well that was simpler than I thought. I didn't realize that cl-defun works that way, but it makes sense. Thanks for reporting.

from bufler.el.

Hi Nick (or should I say No Wayman?

hahaha Whichever you prefer. Always good to have an alias on hand.

When I made that feature, I didn't realize how many places would need to account for it, and how much whack-a-mole I'd have to play when working on other features.

Ah, I see. I've had similar experiences on projects.

Well that was simpler than I thought.

Excellent! Glad to hear it wasn't a difficult fix.

I'm looking forward to integrating bufler into my daily workflow.
Thanks again for the great work and fast turnaround!

from bufler.el.