Failed to byte-compile: Debugger entered--Lisp error: (void-variable this-file)
hrehfeld opened this issue · comments
I don't get a backtrace, only this message. However, I suspect this is caused by https://github.com/raxod502/radian/blob/a1bb6c26d296725fc798e9db935ed409c7c9fb25/emacs/init.el#L68
as it is the only occurrence of that string in my .emacs.d
. In case of an error there, it would error out again on the variable?
Interestingly, when I try to work-around by using:
(unless link-target
(error "Init-file is not a symlink: %s" (or user-init-file
load-file-name
buffer-file-name)))
the compilation error persists: Failed to byte-compile: Debugger entered--Lisp error: (error "Init-file is not a symlink: /home/hrehfeld/.emacs....")
Any way how I can get stack traces for these compilation errors?
Sorry for the extensive delay on response here. You should be able to get a full error message if you run make compile
from the command line.
Hmm, I cannot reproduce, so I'm closing.
was this fixed in c8a2e92?
Ironically, I don't get an error message with make compile
either :(
$ make compile
Loading /home/hrehfeld/.emacs.d/my-init.el (source)...
Loading /home/hrehfeld/.emacs.d/var/recentf-save.el (source)...
Cleaning up the recentf list...
Cleaning up the recentf list...done (0 removed)
Created cite link.
Created citet link.
Created tag link.
Package cl is deprecated
Package rfc2368 is deprecated
Saving all Org buffers...
my-save-some-buffers
Saving all Org buffers... done
make: *** [Makefile:25: compile] Error 1
Interestingly, it seems that some of the greps in byte-compile.bash is failing, as
$ emacs --batch \
--eval "(setq straight-safe-mode t)" \
--load "$HOME/.emacs.d/init.el" \
--funcall radian-batch-byte-compile
works just fine
Hmm, I'm not sure when this was fixed, but that commit does look like a good candidate. The reason you're seeing a compilation error is because the way it's configured currently, any byte-compilation warnings are treated as an error. So, because of the messages like Package cl is deprecated
, the grep statement sees that there was output and fails the build.
This has been rather annoying even to me, despite that having no warnings is a great thing. If you think it would be a good idea, I wouldn't object to making it configurable, so that warnings can be treated as non-fatal.