Please test: macOS dmg for 4.3.92

[Sebastian Schöps]

Dear Mac users,

I created a macOS dmg for Octave 4.3.92 based on homebrew

https://www.dropbox.com/s/fjg8e413seqm6ph/Octave_4-3-92_High_Sierra.dmg?dl=0

with shasum 3c4ddda86c100bfa3ddd092f6d93d810a7cb743e. I applied a few extra patches (#50025 and #49053) and compiled everything on macOS High Sierra (maybe I should use an older version for backward compatibility!?).

Please test and report back.

Best,
Sebastian

[John Swensen-3]

> On May 1, 2018, at 8:00 AM, Sebastian Schöps <[hidden email]> wrote:
>
> Dear Mac users,
>
> I created a macOS dmg for Octave 4.3.92 based on homebrew
>
> https://www.dropbox.com/s/fjg8e413seqm6ph/Octave_4-3-92_High_Sierra.dmg?dl=0
>
> with shasum 3c4ddda86c100bfa3ddd092f6d93d810a7cb743e. I applied a few extra patches (#50025 and #49053) and compiled everything on macOS High Sierra (maybe I should use an older version for backward compatibility!?).
>
> Please test and report back.
>
> Best,
> Sebastian
>
>
>

It is crashing for me. Here are portions of the crash log for the crashing thread.

Crashed Thread:        13  QThread

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Illegal instruction: 4
Termination Reason:    Namespace SIGNAL,

Thread 13 Crashed:: QThread
0   liboctave.5.dylib             0x00000001071627aa void octave_sort<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > >::binarysort<bool (*)(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >*, long long, long long, bool (*)(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)) + 34
1   liboctave.5.dylib             0x0000000107157392 void octave_sort<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > >::sort<bool (*)(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >*, long long, bool (*)(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)) + 378
2   liboctave.5.dylib             0x0000000107592f00 string_vector::sort(bool) + 50
3   liboctinterp.5.dylib           0x0000000106873fdb octave::genpath(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, string_vector const&) + 137
4   liboctinterp.5.dylib           0x00000001066c7d1d octave::environment::init_image_path() + 273
5   liboctinterp.5.dylib           0x00000001068667b1 octave::environment::environment() + 39
6   liboctinterp.5.dylib           0x000000010685ef4c octave::interpreter::interpreter(octave::application*) + 38
7   liboctinterp.5.dylib           0x00000001062089ee octave::application::create_interpreter() + 46
8   liboctgui.3.dylib             0x0000000105f313d5 octave::octave_interpreter::execute() + 43
9   org.qt-project.QtCore         0x0000000108b5a731 QObject::event(QEvent*) + 769
10  org.qt-project.QtWidgets       0x000000010945715d QApplicationPrivate::notify_helper(QObject*, QEvent*) + 269
11  org.qt-project.QtWidgets       0x00000001094584c8 QApplication::notify(QObject*, QEvent*) + 392
12  org.qt-project.QtCore         0x0000000108b30c24 QCoreApplication::notifyInternal2(QObject*, QEvent*) + 164
13  org.qt-project.QtCore         0x0000000108b31df3 QCoreApplicationPrivate::sendPostedEvents(QObject*, int, QThreadData*) + 851
14  org.qt-project.QtCore         0x0000000108b85ab8 QEventDispatcherUNIX::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) + 72
15  org.qt-project.QtCore         0x0000000108b2c03e QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) + 430
16  org.qt-project.QtCore         0x000000010896498c QThread::exec() + 140
17  org.qt-project.QtCore         0x0000000108968a1d 0x10893b000 + 186909
18  libsystem_pthread.dylib       0x00007fff7129c6c1 _pthread_body + 340
19  libsystem_pthread.dylib       0x00007fff7129c56d _pthread_start + 377
20  libsystem_pthread.dylib       0x00007fff7129bc5d thread_start + 13


This same crash occurs whether I run from the launcher or from the command line within the app bundle and with or without GUI. Here are my system specs:

MacOS 10.13.3 (I am waiting for the semester to be over before upgrading)
Homebrew is up-to-date

If you want me to send you the entire crash log, let me know.

John S.

[Sebastian Schöps]

Dear John,

> It is crashing for me. Here are portions of the crash log for the crashing thread.
>
> Crashed Thread:        13  QThread
>
> Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
> Exception Codes:       0x0000000000000001, 0x0000000000000000
> Exception Note:        EXC_CORPSE_NOTIFY
>
> Termination Signal:    Illegal instruction: 4
> Termination Reason:    Namespace SIGNAL,

Thanks for testing! Do you happen to have an older Mac? I should have compiled for "march=core2" but maybe I made a mistake.

Sebastian

[John Swensen-3]

On May 1, 2018, at 11:23 AM, Sebastian Schöps <[hidden email]> wrote:

Dear John,

It is crashing for me. Here are portions of the crash log for the crashing thread.

Crashed Thread:        13  QThread

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Illegal instruction: 4
Termination Reason:    Namespace SIGNAL,

Thanks for testing! Do you happen to have an older Mac? I should have compiled for "march=core2" but maybe I made a mistake.

Sebastian

It is older, but not that old.

MacBook Pro (Retina, 15-inch, Late 2013)

2 GHz Intel Core i7

John S.

[Bradley Kennedy]

On 2018-05-01, at 14:28, John Swensen <[hidden email]> wrote:

On May 1, 2018, at 11:23 AM, Sebastian Schöps <[hidden email]> wrote:

Dear John,

It is crashing for me. Here are portions of the crash log for the crashing thread.

Crashed Thread:        13  QThread

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Illegal instruction: 4
Termination Reason:    Namespace SIGNAL,

Thanks for testing! Do you happen to have an older Mac? I should have compiled for "march=core2" but maybe I made a mistake.

Sebastian

It is older, but not that old.

MacBook Pro (Retina, 15-inch, Late 2013)

2 GHz Intel Core i7

John S.

Hi, I managed to get a crash the first time I ran it, but now the application works without fail in gui/nogui mode.

This crash below occured when double clicking Octave from the applications directory.

The first crash I got was:

Exception Type:        EXC_BAD_ACCESS (SIGSEGV)

Exception Codes:       KERN_INVALID_ADDRESS at 0x000000000000001c

Exception Note:        EXC_CORPSE_NOTIFY

VM Regions Near 0x1c:

--> 

    __TEXT                 000000010ab31000-000000010ab40000 [   60K] r-x/rwx SM=COW  /Applications/Octave.app/Contents/Resources/usr/Cellar/octave/4.3.92/libexec/octave/4.3.92/exec/x86_64-apple-darwin17.5.0/octave-gui

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread

0   org.qt-project.QtWidgets       0x00000001130729e4 QDockWidgetLayout::sizeFromContent(QSize const&, bool) const + 36

1   org.qt-project.QtWidgets       0x000000011307391a 0x112f3d000 + 1272090

2   org.qt-project.QtWidgets       0x0000000113077a6b 0x112f3d000 + 1288811

3   org.qt-project.QtWidgets       0x0000000113077b60 0x112f3d000 + 1289056

4   org.qt-project.QtWidgets       0x0000000113078064 0x112f3d000 + 1290340

…

This is macOS High Sierra v 10.13.1 on a MacBook Air (11-inch Early 2015). More details available on request.

Bradley Kennedy

[hidden email]

[rdzman]

I’m also getting a crash on launch. 2014 MacBook Pro, 10.13.4. Same error as below.

    Ray

On May 1, 2018, at 2:56 PM, Bradley Kennedy <[hidden email]> wrote:

On 2018-05-01, at 14:28, John Swensen <[hidden email]> wrote:

On May 1, 2018, at 11:23 AM, Sebastian Schöps <[hidden email]> wrote:

Dear John,

It is crashing for me. Here are portions of the crash log for the crashing thread.

Crashed Thread:        13  QThread

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Illegal instruction: 4
Termination Reason:    Namespace SIGNAL,

Thanks for testing! Do you happen to have an older Mac? I should have compiled for "march=core2" but maybe I made a mistake.

Sebastian

It is older, but not that old.

MacBook Pro (Retina, 15-inch, Late 2013)

2 GHz Intel Core i7

John S.

Hi, I managed to get a crash the first time I ran it, but now the application works without fail in gui/nogui mode.

This crash below occured when double clicking Octave from the applications directory.

The first crash I got was:

Exception Type:        EXC_BAD_ACCESS (SIGSEGV)

Exception Codes:       KERN_INVALID_ADDRESS at 0x000000000000001c

Exception Note:        EXC_CORPSE_NOTIFY

VM Regions Near 0x1c:

--> 

    __TEXT                 000000010ab31000-000000010ab40000 [   60K] r-x/rwx SM=COW  /Applications/Octave.app/Contents/Resources/usr/Cellar/octave/4.3.92/libexec/octave/4.3.92/exec/x86_64-apple-darwin17.5.0/octave-gui

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread

0   org.qt-project.QtWidgets       0x00000001130729e4 QDockWidgetLayout::sizeFromContent(QSize const&, bool) const + 36

1   org.qt-project.QtWidgets       0x000000011307391a 0x112f3d000 + 1272090

2   org.qt-project.QtWidgets       0x0000000113077a6b 0x112f3d000 + 1288811

3   org.qt-project.QtWidgets       0x0000000113077b60 0x112f3d000 + 1289056

4   org.qt-project.QtWidgets       0x0000000113078064 0x112f3d000 + 1290340

…

This is macOS High Sierra v 10.13.1 on a MacBook Air (11-inch Early 2015). More details available on request.

Bradley Kennedy

[hidden email]

[c.-2]

> On 1 May 2018, at 17:00, Sebastian Schöps <[hidden email]> wrote:
>
> Dear Mac users,
>
> I created a macOS dmg for Octave 4.3.92 based on homebrew
>
> https://www.dropbox.com/s/fjg8e413seqm6ph/Octave_4-3-92_High_Sierra.dmg?dl=0
>
> with shasum 3c4ddda86c100bfa3ddd092f6d93d810a7cb743e. I applied a few extra patches (#50025 and #49053) and compiled everything on macOS High Sierra (maybe I should use an older version for backward compatibility!?).
>
> Please test and report back.
>
> Best,
> Sebastian

Hi,

It does launch without crash for me on macos High Sierra 10.13.4 on a MacBook Pro 13-inch, 2016.

But __run_test_suite__ returns a few FAIL that I don't see on a self-compiled 4.4.0, i.e.:

Summary:

  PASS                            14896
  FAIL                               15
  XFAIL (reported bug)               54
  SKIP (missing feature)             77
  SKIP (run-time condition)          28

13 of the 15 FAIL are in "vecnorm.m" which passes all test in my self-compiled 4.4.0.

Thanks for your work!
c.

[c.-2]

> On 2 May 2018, at 13:42, c. <[hidden email]> wrote:
>
> 13 of the 15 FAIL are in "vecnorm.m" which passes all test in my self-compiled 4.4.0.

actually it seems this problem is related to a version of vecnorm in the nurbs package as I have "pkg load nurbs" in my .octaverc

>> pkg unload nurbs
>> test vecnorm
PASSES 13 out of 13 tests

So this is not an issue with the macos binary.

Rafael, may I suggest we change name to the vecnorm function in the nurbs package?
Also Matlab has a built-in vecnorm function that has different functionality an I don't think it is a good idea to shadow it with the one in nurbs ...

c.

[Sebastian Schöps]

Dear John, dear Bradley,

> It is crashing for me. Here are portions of the crash log for the crashing thread.

I made a new dmg. It is now based on Octave 4.4 and compiled using Yosemite in a Virtual Machine...

https://www.dropbox.com/s/ywmszs0kwzhaxzn/Octave_44_Yosemite.dmg?dl=0

Is it still crashing for you?

Best
Sebastian

[Rafael Vazquez-2]

On 02. 05. 18 13:53, c. wrote:

>
>> On 2 May 2018, at 13:42, c. <[hidden email]> wrote:
>>
>> 13 of the 15 FAIL are in "vecnorm.m" which passes all test in my self-compiled 4.4.0.
> actually it seems this problem is related to a version of vecnorm in the nurbs package as I have "pkg load nurbs" in my .octaverc
>
>>> pkg unload nurbs
>>> test vecnorm
> PASSES 13 out of 13 tests
>
> So this is not an issue with the macos binary.
>
> Rafael, may I suggest we change name to the vecnorm function in the nurbs package?
> Also Matlab has a built-in vecnorm function that has different functionality an I don't think it is a good idea to shadow it with the one in nurbs ...
>
> c.
>
>

I agree. The Matlab built-in vecnorm probably did not exist when the
package was created.

I suggest to change the name to vecnormalize. If you agree, I will do
all the necessary changes.

Rafa

[c.-2]

> On 2 May 2018, at 14:01, Rafael Vazquez <[hidden email]> wrote:
>
> I agree. The Matlab built-in vecnorm probably did not exist when the package was created.
>
> I suggest to change the name to vecnormalize. If you agree, I will do all the necessary changes.
Sounds good. Please don't make the release yet though, as I think we need to make more changes to nurbs to make it compatible with 4.4.0

> Rafa
c.

[rdzman]

This one did not crash on launch. However, when I do __run_test_suite__ it runs the tests, then pops up this dialog which I can’t get rid of. Clicking the buttons does nothing. Hitting Enter or Esc does nothing. I can still use the GUI menus to show the Command window and see that there were 2 FAIL’s. Any ideas why this dialog comes up … and why I can’t get rid of it?

   Ray

On May 2, 2018, at 7:59 AM, Sebastian Schöps <[hidden email]> wrote:

Dear John, dear Bradley,

It is crashing for me. Here are portions of the crash log for the crashing thread.

I made a new dmg. It is now based on Octave 4.4 and compiled using Yosemite in a Virtual Machine...

https://www.dropbox.com/s/ywmszs0kwzhaxzn/Octave_44_Yosemite.dmg?dl=0

Is it still crashing for you?

Best
Sebastian

[Mike Miller-4]

On Wed, May 02, 2018 at 14:51:02 -0400, Ray Zimmerman wrote:
> This one did not crash on launch. However, when I do
> __run_test_suite__ it runs the tests, then pops up this dialog which I
> can’t get rid of. Clicking the buttons does nothing. Hitting Enter or
> Esc does nothing. I can still use the GUI menus to show the Command
> window and see that there were 2 FAIL’s. Any ideas why this dialog
> comes up … and why I can’t get rid of it?

I know why the dialog appears (https://savannah.gnu.org/bugs/?53539).

I don't know why you can't close it, it can be closed on non-macOS
systems.

--
mike

[Sebastian Schöps]

Mike Miller-4 wrote

> On Wed, May 02, 2018 at 14:51:02 -0400, Ray Zimmerman wrote:
>> This one did not crash on launch. However, when I do
>> __run_test_suite__ it runs the tests, then pops up this dialog which I
>> can’t get rid of. Clicking the buttons does nothing. Hitting Enter or
>> Esc does nothing. I can still use the GUI menus to show the Command
>> window and see that there were 2 FAIL’s. Any ideas why this dialog
>> comes up … and why I can’t get rid of it?
>
> I know why the dialog appears (https://savannah.gnu.org/bugs/?53539).
>
> I don't know why you can't close it, it can be closed on non-macOS
> systems.

Strange, I do not see that message at all!?

BTW: I have one fail in tar.m, which I'd say is noncritical.

Sebastian




--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

[Mike Miller-4]

On Wed, May 02, 2018 at 13:10:34 -0700, Sebastian Schöps wrote:
> BTW: I have one fail in tar.m, which I'd say is noncritical.

I know, I have that filed as https://savannah.gnu.org/bugs/?53695, which
I plan to look at fixing eventually.

If you're able to include GNU tar in your app bundle, and rename it to
tar and have it override the system tar, that might help your users for
now.

--
mike

[John Swensen-3]

On May 2, 2018, at 4:59 AM, Sebastian Schöps <[hidden email]> wrote:

>
> Dear John, dear Bradley,
>
>> It is crashing for me. Here are portions of the crash log for the crashing thread.
>
> I made a new dmg. It is now based on Octave 4.4 and compiled using Yosemite in a Virtual Machine...
>
> https://www.dropbox.com/s/ywmszs0kwzhaxzn/Octave_44_Yosemite.dmg?dl=0
>
> Is it still crashing for you?
>
> Best
> Sebastian

This one works for me also, but I get the same dialog (but with a different file) when running the tests. I got one FAIL and a bunch of XFAIL, but because that dialog was blocking the UI, I couldn’t scroll up to see where the fail happened.

John S.

[Sebastian Schöps]

Mike Miller-4 wrote
> On Wed, May 02, 2018 at 13:10:34 -0700, Sebastian Schöps wrote:
>> BTW: I have one fail in tar.m, which I'd say is noncritical.
>
> I know, I have that filed as https://savannah.gnu.org/bugs/?53695, which
> I plan to look at fixing eventually.
>
> If you're able to include GNU tar in your app bundle, and rename it to
> tar and have it override the system tar, that might help your users for
> now.

I just figured that out myself and came back here to proudly report and now
I must read that you knew it all along :) However, I did exactly as you
said.

Thanks,
Sebastian

P.S.: I made a comment on #53695.



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

[Sebastian Schöps]

John Swensen-3 wrote
> This one works for me also, but I get the same dialog (but with a
> different file) when running the tests. I got one FAIL and a bunch of
> XFAIL, but because that dialog was blocking the UI, I couldn’t scroll up
> to see where the fail happened.

There is a new build

https://www.dropbox.com/s/1fpk4xagqed81om/Octave_44_Yosemite.dmg?dl=0

that should fix the remaining FAIL (now using GNU tar instead of BSD tar).

I have no clue how to get rid of the dialog but as this is a known bug, I'd
say we live with it for now. Luckily, it seems only to affect the tests.

Best,
Sebastian



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

[rdzman]

With this one, I was able to dismiss the dialog. But I’m still getting 3 FAILs when running with the GUI, 2 FAILS without, as listed below. Plus, I’m getting a lot of repetitions of the octave-gui fails assertion you see. Again, this is on a mid-2014 MacBook Pro running High Sierra 10.13.4.

    Ray

———————————————————————————

With GUI …

  miscellaneous/dir.m ......................................... PASS      2/3

                                                                  FAIL    1

  miscellaneous/tar.m .........................................2018-05-03 13:08 octave-gui[7315] (FSEvents.framework) FSEventStreamFlushSync(): failed assertion '(SInt64)last_id > 0LL'

a dir2
a dir2/file2
a file1
2018-05-03 13:08 octave-gui[7315] (FSEvents.framework) FSEventStreamFlushSync(): failed assertion '(SInt64)last_id > 0LL'

2018-05-03 13:08 octave-gui[7315] (FSEvents.framework) FSEventStreamFlushSync(): failed assertion '(SInt64)last_id > 0LL'

 PASS      5/6
                                                                  FAIL    1

  plot/appearance/legend.m .................................... PASS      4/5
                                                                  FAIL    1

Summary:

  PASS                            14908

  FAIL                                3

  XFAIL (reported bug)               54

  SKIP (missing feature)             77

  SKIP (run-time condition)          28

———————————————————————————

Without GUI …

  miscellaneous/dir.m ......................................... PASS      2/3   
                                                                  FAIL    1

  miscellaneous/tar.m .........................................a dir2
a dir2/file2
a file1
 PASS      5/6   
                                                                  FAIL    1

  plot/appearance/legend.m .................................... PASS      5/5   

Summary:

  PASS                            14909
  FAIL                                2
  XFAIL (reported bug)               54
  SKIP (missing feature)             77
  SKIP (run-time condition)          28

On May 3, 2018, at 7:23 AM, Sebastian Schöps <[hidden email]> wrote:

John Swensen-3 wrote

This one works for me also, but I get the same dialog (but with a
different file) when running the tests. I got one FAIL and a bunch of
XFAIL, but because that dialog was blocking the UI, I couldn’t scroll up
to see where the fail happened.

There is a new build

https://www.dropbox.com/s/1fpk4xagqed81om/Octave_44_Yosemite.dmg?dl=0

that should fix the remaining FAIL (now using GNU tar instead of BSD tar).

I have no clue how to get rid of the dialog but as this is a known bug, I'd
say we live with it for now. Luckily, it seems only to affect the tests.

Best,
Sebastian



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

[Simone Deparis]

Dear Sebastian,

if I double click on the icon, nothing appens.

 From Terminal:
/Applications/Octave.app/Contents/Resources/usr/bin/octave-cli
Illegal instruction: 4

same with octave (without -cli)

I have High Sierra 10.13.4

Best
Simone


On 02.05.18 14:10, c. wrote:

>
>
>> On 2 May 2018, at 14:01, Rafael Vazquez <[hidden email]> wrote:
>>
>> I agree. The Matlab built-in vecnorm probably did not exist when the package was created.
>>
>> I suggest to change the name to vecnormalize. If you agree, I will do all the necessary changes.
> Sounds good. Please don't make the release yet though, as I think we need to make more changes to nurbs to make it compatible with 4.4.0
>
>> Rafa
> c.
>
>

--
_____________________________________________________________________

   Simone Deparis, PhD, MER
   Deputy Director SMA
   MA A2 383
   SMA - SB - EPFL
   Station 8
   CH - 1015 Lausanne
   Switzerland

   phone :  [phone]             mailto:[hidden email]
                                  http://people.epfl.ch/simone.deparis
_____________________________________________________________________