Issue #90 reappears after commit fcc513d6

[dalugm]

Seems like #90 was solved by adding (load-file file), but commit fcc513d removed that line

My emacs config is also multi-file based, now I have the same problem as #90

How can I avoid error when running make native, it reminds me

../../init.el:300:11: Error: Cannot open load file: No such file or directory, init-modeline

[tarsius]

I have released v4.4.0 without any further changes or reverts in this area, but that doesn't mean the current behavior is set in stone.

I've also just noticed that you have updated your Emacs configuration a few days, potentially to resolve this issue differently. You have updated to the latest convention of doing Borg initialization in early-init.el.

Have you also set the environment variable EMACS_Q_ARG to -q (instead of the default -Q)? This should make the (load-file self) unnecessary, I believe.

[dalugm]

Yes, I think the current issue is that when Borg compiles my init.el, it doesn't know that lisp/ is also in my load-path. If I put all my configuration directly into init.el, this problem would be avoided (though I'm not sure whether putting everything into init.el is a good idea — I already have over 2,000 lines in my init.el, and moving the rest of the code from my lisp/ files into it would probably balloon it to around 3,000 lines...)

Have you also set the environment variable EMACS_Q_ARG to -q (instead of the default -Q)? This should make the (load-file self) unnecessary, I believe.

Emmm... I've never manually set EMACS_Q_ARG before, but I tried setting it to -q, and the issue still persists.

EMACS_Q_ARG=-q
make quick

# result
../../init.el:548:11: Error: Cannot open load file: No such file or directory, init-modeline

[tarsius]

and moving the rest of the code from my lisp/ files into it would probably balloon it to around 3,000 lines...)

Yeah, you wouldn't want to do that. Let's assume for a moment that making some small changes to your configuration doesn't do the trick, and that I refuse to make the-necessary-change™ to Borg, then the reasonable thing for you to do, would be to patch Borg, not to completely restructure your configuration.

Emmm... I've never manually set EMACS_Q_ARG before, but I tried setting it to -q, and the issue still persists.

Well, you would have to also move that (add-to-list 'load-path (expand-file-name "lisp" user-emacs-directory) to early-init.el.

[dalugm]

Emmm... Thanks, I’m still quite lazy at this -- I always hope others will just point me directly to the answer. But after looking closely at the actual code, I realized that make init-build actually calls the function borg-batch-rebuild-init, and within that function, byte-recompile-file is what does the heavy lifting. Now I have a thoughrt of how to fix my code

My issue was that I directly wrote this in init.el:

(add-to-list 'load-path (expand-file-name "lisp" user-emacs-directory))

(require 'init-xxx)

This only takes effect at evaluation time, not at compile time. To make it work during compilation as well, I just need to wrap it in eval-and-compile:

(eval-and-compile
  (add-to-list 'load-path (expand-file-name "lisp" user-emacs-directory))

  (require 'init-xxx))

With this change, I don’t need to modify EMACS_Q_ARGS, nor do I need to move the load-path modification into early-init.el. The required files will be found during both compilation and evaluation.

I believe this might be the most appropriate solution for me -- but since I only have a superficial understanding of Elisp, please feel free to let me know if there’s anything wrong with this approach ;)

[tarsius]

Great! Sounds good too.