Straight.el: Breaks eval-after-load with string argument
What's wrong
While using straight.el, eval-after-load with string argument does not get called.
Directions to reproduce
init.el
;;; bootstrap straight.el
;;; .......
(eval-after-load "evil" '(progn (message "loaded evil string")))
(eval-after-load 'evil '(progn (message "loaded evil symbol")))
(straight-use-package 'evil)
(require 'evil)
(evil-mode t)
expected behavior
Both "loaded evil string" and "loaded evil symbol" should be printed.
actual behavior
Only "loaded evil symbol" is printed.
Without straight.el
(eval-after-load "evil" '(progn (message "loaded evil string")))
(eval-after-load 'evil '(progn (message "loaded evil symbol")))
(add-to-list 'load-path "~/.emacs.d/evil/")
(require 'evil)
(evil-mode t)
This would produce the expected behavior
Version information
- Emacs version: 28 native-comp branch
- Operating system: linux
blahgeek
All 10 comments
Your example works for me:
Bootstrapping straight.el...done
Rebuilding all packages due to build cache schema change
Looking for gnu-elpa-mirror recipe → Cloning melpa...done
Looking for emacsmirror-mirror recipe → Cloning gnu-elpa-mirror...done
Looking for straight recipe → Cloning emacsmirror-mirror...done
Building straight...done
Cloning evil...done
Building evil...
Building evil → Cloning undo-tree...done
Building evil → Building undo-tree...done
Building evil → Cloning goto-chg...done
Building evil → Building goto-chg...done
Building evil...done
loaded evil string
loaded evil symbol
For information about GNU Emacs and the GNU system, type C-h C-a.
Running GNU Emacs 28.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.21, cairo version 1.17.3) of 2020-08-10 on Manjaro Linux.
raxod502
on 24 Aug 2020
Yes I just found that it works in the first launch. But not on the second launch. Also, removing build-cache.el seems help.
blahgeek
on 24 Aug 2020
On second launch, I get:
loaded evil string
loaded evil symbol
For information about GNU Emacs and the GNU system, type C-h C-a.
Did you get unlucky and get a temporary regression in the commit of Emacs master that you built?
raxod502
on 25 Aug 2020
That's odd. Thanks for your response, I'll check another version of emacs when I get time.
blahgeek
on 25 Aug 2020
It seems that it only happens in native-comp branch of emacs.
blahgeek
on 29 Aug 2020
Can you reproduce the error without straight.el by adding Evil to the load-path and manually invoking the native compiler?
raxod502
on 31 Aug 2020
No. I cannot reproduce this with (native-compile-async "~/.emacs.d/evil/" 'recursively 'late) without straight.el.
Also, with straight.el, setting straight-disable-native-compilation and/or straight-disable-byte-compilation does not make a difference.
However, setting straight-disable-native-compilation to t and straight-cache-autoloads to nil fixes this.
Maybe not directly related: according to https://akrl.sdf.org/gccemacs.html#orgd0c37e9, explicitly calling native compilation seems not required anymore.
blahgeek
on 13 Sep 2020
setting straight-disable-native-compilation and/or straight-disable-byte-compilation does not make a difference
See #569.
setting both straight-disable-native-compilation and straight-cache-autoloads to t
That is distinctly horrifying, as straight-cache-autoloads has a non-nil value by default.
raxod502
on 20 Sep 2020
@raxod502 So sorry, typo, I meant setting straight-cache-autoloads to nil. Last comment was editted.
blahgeek
on 20 Sep 2020
This issue is resolved after upgrading emacs-native-comp from 28.0.50.143324 to 28.0.50.144078 (both from archlinux AUR). So I may have indeed get unlucky and get a temporary regression from emacs master :joy: I'm closing this issue now, thanks again for your help.
blahgeek
on 29 Sep 2020
Related issues
raxod502
·
3Comments
rsbowman
·
3Comments
progfolio
·
3Comments
dertuxmalwieder
·
3Comments
aspiers
·
4Comments