origin/features/native-comp: Where do jitted .eln files go?

origin/features/native-comp: Where do jitted .eln files go?

[T.V Raman]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=gb18030, Size: 736 bytes --]


Hi Andrea and others on the native-branch:

I successfully built the native-comp branch and ran it --- as elisp libs
get loaded from my elpa/ directory etc, I see  Emacs saying that it's
native compiling those files -- however I have no idea where it places
those (I can see the .eln files in my Emacs build for the .el files that
come with Emacs).

Additional question: If I wanted to native-compile a given elisp
package, what elisp function should I call to do the native compile?

Finally: It wasn't obvious that doing the native compile requireds the
--with-zip flag to configure.

Thanks Andrea for the awesome work to get the native-comp branch to such
a mature state!
-- 
7©4Id: kg:/m/0285kf1  •0Ü87©4

Re: origin/features/native-comp: Where do jitted .eln files go?

[Andrea Corallo via Emacs development discussions.]

"T.V Raman" <raman@google.com> writes:

> Hi Andrea and others on the native-branch:
>
> I successfully built the native-comp branch and ran it --- as elisp libs
> get loaded from my elpa/ directory etc, I see  Emacs saying that it's
> native compiling those files -- however I have no idea where it places
> those (I can see the .eln files in my Emacs build for the .el files that
> come with Emacs).
>
> Additional question: If I wanted to native-compile a given elisp
> package, what elisp function should I call to do the native compile?

Hi Raman,

I believe the last update here https://akrl.sdf.org should answer your
questions on where the .eln are :)

The compilation for packages is supposed to happen automatically, but if
you really want to invoke the native compilation manually to compile AoT
a package you can still invoke something like:

(native-compile-async "~/.emacs.d/elpa/etcetc" 4 t)

> Finally: It wasn't obvious that doing the native compile requireds the
> --with-zip flag to configure.

Yeah that's a dependency introduced very recently (this weekend :), we
may remove it depending on what's the final strategy for hashing the the
eln filenames we decide to go for.

> Thanks Andrea for the awesome work to get the native-comp branch to such
> a mature state!

Glad you enjoy it!

Thanks

  Andrea

-- 
akrl@sdf.org

Possible Race condition? Re: origin/features/native-comp: Where do jitted .eln files go?

[T.V Raman]

Hi Andrea,

In my emacs startup, I load dired-x. In my customization settings, I
have dired-omit-mode in my dired-mode-hook.

Works fine with plain emacs, but throws this error in *Messages*
(starting with --debug-init does not produce a backtrace.

Symbol’s function definition is void: dired-omit-mode

Suspect some kind of race condition.
f
-- 
♉Id: kg:/m/0285kf1  🦮♉

--
♉Id: kg:/m/0285kf1  🦮♉

Re: Possible Race condition? Re: origin/features/native-comp: Where do jitted .eln files go?

[Andrea Corallo via Emacs development discussions.]

"T.V Raman" <raman@google.com> writes:

> Hi Andrea,
>
> In my emacs startup, I load dired-x. In my customization settings, I
> have dired-omit-mode in my dired-mode-hook.
>
> Works fine with plain emacs, but throws this error in *Messages*
> (starting with --debug-init does not produce a backtrace.
>
> Symbol’s function definition is void: dired-omit-mode
>
> Suspect some kind of race condition.
> f

Hi Raman,

I'd bet more on some other kind of bug.  If you could isolate it and
open a bug with a reproducer would be appretiated.

Thanks!

  Andrea
  
-- 
akrl@sdf.org

Re: Possible Race condition? Re: origin/features/native-comp: Where do jitted .eln files go?

[T.V Raman]

Any tips on how to isolate / produce  a bug trace? Note that
--debug-init did not produce a backgtrace Andrea Corallo writes:
 > "T.V Raman" <raman@google.com> writes:
 > 
 > > Hi Andrea,
 > >
 > > In my emacs startup, I load dired-x. In my customization settings, I
 > > have dired-omit-mode in my dired-mode-hook.
 > >
 > > Works fine with plain emacs, but throws this error in *Messages*
 > > (starting with --debug-init does not produce a backtrace.
 > >
 > > Symbol’s function definition is void: dired-omit-mode
 > >
 > > Suspect some kind of race condition.
 > > f
 > 
 > Hi Raman,
 > 
 > I'd bet more on some other kind of bug.  If you could isolate it and
 > open a bug with a reproducer would be appretiated.
 > 
 > Thanks!
 > 
 >   Andrea
 >   
 > -- 
 > akrl@sdf.org

-- 
♉Id: kg:/m/0285kf1  🦮♉

--
♉Id: kg:/m/0285kf1  🦮♉

Re: Possible Race condition? Re: origin/features/native-comp: Where do jitted .eln files go?

[Andrea Corallo via Emacs development discussions.]

"T.V Raman" <raman@google.com> writes:

> Any tips on how to isolate / produce  a bug trace? Note that
> --debug-init did not produce a backgtrace Andrea Corallo writes:

Mmh sorry I've no smart suggestion here as I've never seen anything like
that.  I'd first look carefully if by chance there's something
interesting in *Messages* an then I'd probably try to understated what is
causing the issue in the .emacs commenting out stuffs by bisection.

If you get a minimal reproducer even if there's no backtrace I should be
able to reproduce it and look into it with gdb in the worst case.

  Andrea

-- 
akrl@sdf.org

Re: origin/features/native-comp: Where do jitted .eln files go?

[T.V Raman]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=gb18030, Size: 944 bytes --]

Andrea Corallo <akrl@sdf.org> writes:


Thanks!
In the meantime:

I rebuilt the native-branch, blew away my .emacs.d/eln-cache, eliminated
 the error from dired-x (got rid of with-eval-after-load for that use
 case) and now 99% of emacspeak appears to work reliably in native-emacs
 --- still hunting down corner cases where it might be
 failing. Incidentally I got a segfault when I hit tab in the minibuffer
 while in vm (mail reader) haven't tracked it down yet > "T.V Raman" <raman@google.com> writes:
>
>> Hi Andrea,
>> Any plans to introduce a native equivalent of batch-byte-compile ---
>> might be useful in cases where  load order matters e.g. with heavy use
>> of defadvice 
>
> Hi Raman,
>
> We have `batch-native-compile',  ATM this is producing only the .eln
> files so also batch-byte-compile must be invoked but it should do the
> job.
>
> Regards
>
>   Andrea

-- 
7©4 Id: kg:/m/0285kf1  •0Ü8

Re: origin/features/native-comp: Where do jitted .eln files go?

[Andrea Corallo via Emacs development discussions.]

"T.V Raman" <raman@google.com> writes:

> I rebuilt the native-branch, blew away my .emacs.d/eln-cache, eliminated
>  the error from dired-x (got rid of with-eval-after-load for that use
>  case) and now 99% of emacspeak appears to work reliably in
>  native-emacs

Sounds great!

You should also be able to use with-eval-after-load now as its fixed
(bug#43089).

>  --- still hunting down corner cases where it might be
>  failing. Incidentally I got a segfault when I hit tab in the minibuffer
>  while in vm (mail reader) haven't tracked it down yet > "T.V Raman"
>  <raman@google.com>

I don't see a segfault on my system since a long time, if you manage to
make it reproducible would be certainly interesting.

Thanks

  Andrea