Hi there, I've been trying out the nativecomp branch and I use straight.el as my package manager. One of the things that it does is throw when it detects that its own compiled bytecode has changed. It does this from within straight.el: (eval `(unless (equal (emacs-version) ,(eval-when-compile (emacs-version))) (throw 'emacs-version-changed nil))) https://github.com/raxod502/straight.el/blob/2d407bccd9378f1d5218f8ba2ae85c6be73fbaf1/straight.el#L45-L49 It then catches within bootstrap.el, which loads straight.el: (catch 'emacs-version-changed ;; straight.el has a fun hack that throws ;; `emacs-version-changed' if the version of Emacs has changed ;; since the last time it was byte-compiled. This prevents us ;; from accidentally loading invalid byte-code, hopefully. (load (expand-file-name (concat straight.el "c") default-directory) nil 'nomessage 'nosuffix) (setq emacs-version-changed nil)) https://github.com/raxod502/straight.el/blob/2d407bccd9378f1d5218f8ba2ae85c6be73fbaf1/bootstrap.el#L47-L54 Unfortunately, this doesn't work with nativecomp. It claims that there's no matching catch when a throw actually happens. Warning (initialization): An error occurred while loading ‘/Users/aaronjensen/.emacs.d/early-init.el’: No catch for tag: emacs-version-changed, nil To reproduce, you should be able to use the bootstrap code: https://github.com/raxod502/straight.el#bootstrapping-straightel Then recompile emacs (which changes the value of emacs-version) and run again. I tried narrowing it down to just a try/catch in two different files and that worked, so there must be something more peculiar going on here. Aaron
Aaron Jensen <aaronjensen@gmail.com> writes:
> Hi there,
>
> I've been trying out the nativecomp branch and I use straight.el as my
> package manager.
>
> One of the things that it does is throw when it detects that its own
> compiled bytecode has changed. It does this from within straight.el:
>
> (eval
> `(unless (equal
> (emacs-version)
> ,(eval-when-compile (emacs-version)))
> (throw 'emacs-version-changed nil)))
>
> https://github.com/raxod502/straight.el/blob/2d407bccd9378f1d5218f8ba2ae85c6be73fbaf1/straight.el#L45-L49
>
> It then catches within bootstrap.el, which loads straight.el:
>
> (catch 'emacs-version-changed
> ;; straight.el has a fun hack that throws
> ;; `emacs-version-changed' if the version of Emacs has changed
> ;; since the last time it was byte-compiled. This prevents us
> ;; from accidentally loading invalid byte-code, hopefully.
> (load (expand-file-name (concat straight.el "c")
> default-directory)
> nil 'nomessage 'nosuffix)
> (setq emacs-version-changed nil))
>
> https://github.com/raxod502/straight.el/blob/2d407bccd9378f1d5218f8ba2ae85c6be73fbaf1/bootstrap.el#L47-L54
>
> Unfortunately, this doesn't work with nativecomp. It claims that
> there's no matching catch when a throw actually happens.
>
> Warning (initialization): An error occurred while loading
> ‘/Users/aaronjensen/.emacs.d/early-init.el’:
>
> No catch for tag: emacs-version-changed, nil
>
>
> To reproduce, you should be able to use the bootstrap code:
> https://github.com/raxod502/straight.el#bootstrapping-straightel
>
> Then recompile emacs (which changes the value of emacs-version) and run again.
>
> I tried narrowing it down to just a try/catch in two different files
> and that worked, so there must be something more peculiar going on
> here.
Hi Aaron,
is there a corresponding issue open in straight? Perhaps straight
developers can narrow down the issue and suggest a minimal reproducer in
case this is a real issue on our side.
Thanks
Andrea
On Thu, Feb 18, 2021 at 2:06 AM Andrea Corallo <akrl@sdf.org> wrote: > > Hi Aaron, > > is there a corresponding issue open in straight? Perhaps straight > developers can narrow down the issue and suggest a minimal reproducer in > case this is a real issue on our side. Reported here: https://github.com/raxod502/straight.el/issues/697 I believe the issue is that in the case that the version changes, the throw works and the catch works. However, immediately after that, they byte-compile straight.el and then attempt to load the elc. However, because there's an eln, the eln is preferred and it throws again because it wasn't rebuilt. If I rebuild the eln as well, it works as expected. https://github.com/raxod502/straight.el/blob/2d407bccd9378f1d5218f8ba2ae85c6be73fbaf1/bootstrap.el#L62-L65 Would it make sense to somehow invalidate the eln if the elc is recompiled on the emacs side? Or should straight detect native-compile and run it as well as byte-compile-file in this scenario? Thanks, Aaron
Aaron Jensen <aaronjensen@gmail.com> writes:
> Would it make sense to somehow invalidate the eln if the elc is
> recompiled on the emacs side? Or should straight detect native-compile
> and run it as well as byte-compile-file in this scenario?
It should be already the case. If you find a reproducer that highlight
the issue or there's evidence of this please file a bug in the bug
tracker so we can work on that.
Thanks!
Andrea
On Thu, Feb 18, 2021 at 3:44 AM Andrea Corallo <akrl@sdf.org> wrote: > > It should be already the case. If you find a reproducer that highlight > the issue or there's evidence of this please file a bug in the bug > tracker so we can work on that. Done, https://debbugs.gnu.org/cgi/bugreport.cgi?bug=46617 Thanks, Aaron
Aaron Jensen <aaronjensen@gmail.com> writes:
> On Thu, Feb 18, 2021 at 3:44 AM Andrea Corallo <akrl@sdf.org> wrote:
>>
>> It should be already the case. If you find a reproducer that highlight
>> the issue or there's evidence of this please file a bug in the bug
>> tracker so we can work on that.
>
> Done, https://debbugs.gnu.org/cgi/bugreport.cgi?bug=46617
>
> Thanks,
>
> Aaron
Thanks Aaron, will look at.
Andrea