Race condition on make?

Race condition on make?

[Juanma Barranquero]

[-- Attachment #1: Type: text/plain, Size: 332 bytes --]

$ make
  ELC+ELN  ../lisp/simple.elc
  ELC+ELN  ../lisp/minibuffer.elc
Reloading Rstealloading se tsailmpele .eslimp
le.el
LoadinLg d:o/ading
d:/Devel/emacs/repo/truDnekvel/emacs/repo/trunk/lisp/l/isp/simple.el
(source)...simp
le.el (source)...
Reloading stale minibuffer.el

Is this the result of a race condition with a -jN make?

[-- Attachment #2: Type: text/html, Size: 457 bytes --]

Re: Race condition on make?

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Mon, 14 Nov 2022 09:53:15 +0100
> 
> $ make
>   ELC+ELN  ../lisp/simple.elc
>   ELC+ELN  ../lisp/minibuffer.elc
> Reloading Rstealloading se tsailmpele .eslimp
> le.el
> LoadinLg d:o/ading d:/Devel/emacs/repo/truDnekvel/emacs/repo/trunk/lisp/l/isp/simple.el (source)...simp
> le.el (source)...
> Reloading stale minibuffer.el
> 
> Is this the result of a race condition with a -jN make?

Where do you see a race condition?

Re: Race condition on make?

[Juanma Barranquero]

[-- Attachment #1: Type: text/plain, Size: 98 bytes --]

I see the output of several parallel instances of make (run with a single
"make -j4") intermixed.

[-- Attachment #2: Type: text/html, Size: 199 bytes --]

Re: Race condition on make?

[Gerd Möllmann]

I think you are seeing bug#58653 (not a bug in Emacs, I guess).

I'm now using MAKEFLAGS="-j8 --output-sync=line", with gmake 4.4.  Older 
versions, for instance 3.81 which comes with Xcode, don't seem to 
support --output-sync.  If not syncing is a feature, I'm kind of in 
doubt, but hey...

Re: Race condition on make?

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Tue, 15 Nov 2022 05:23:56 +0100
> Cc: emacs-devel@gnu.org
> 
> I see the output of several parallel instances of make (run with a single "make -j4") intermixed.

That's because 4 programs are writing at the same time to the same
device.  The OS in most cases succeeds in separating the outputs, but
that is not guaranteed.

Re: Race condition on make?

[Juanma Barranquero]

[-- Attachment #1: Type: text/plain, Size: 60 bytes --]

It's GNU make 4.3, so I'll try --output-sync=line.

Thanks.

[-- Attachment #2: Type: text/html, Size: 165 bytes --]

Re: Race condition on make?

[Juanma Barranquero]

[-- Attachment #1: Type: text/plain, Size: 338 bytes --]

On Tue, Nov 15, 2022 at 2:49 PM Eli Zaretskii <eliz@gnu.org> wrote:

> That's because 4 programs are writing at the same time to the same
> device.  The OS in most cases succeeds in separating the outputs, but
> that is not guaranteed.

Ok, though I had never seen such a thing before in many, many emacs
compilations and bootstrappings.

[-- Attachment #2: Type: text/html, Size: 521 bytes --]

Re: Race condition on make?

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Tue, 15 Nov 2022 20:00:27 +0100
> Cc: emacs-devel@gnu.org
> 
> On Tue, Nov 15, 2022 at 2:49 PM Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > That's because 4 programs are writing at the same time to the same
> > device.  The OS in most cases succeeds in separating the outputs, but
> > that is not guaranteed.
> 
> Ok, though I had never seen such a thing before in many, many emacs compilations and bootstrappings.

It's very rare on MS-Windows, at least IME.