4.2.2 release

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

4.2.2 release

John W. Eaton
Administrator
I'd like to make the 4.2.2 release now.

One of the most visible issues it should fix is the crash on exit.  My
build of the current stable sources does appear to work properly and teh
crash no longer happens.  Does anyone else still experience the crash?

Thanks,

jwe


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Daniel Sebald
On 02/15/2018 10:37 AM, John W. Eaton wrote:
> I'd like to make the 4.2.2 release now.
>
> One of the most visible issues it should fix is the crash on exit.  My
> build of the current stable sources does appear to work properly and teh
> crash no longer happens.  Does anyone else still experience the crash?
>
> Thanks,

Not related to the exit crash, here's another one to consider before
release:

http://savannah.gnu.org/bugs/?53137

I can't say this is a problem, and I can't point to any manifestation of
the color table being accessed prematurely.  However, in doing
development here, every once in a while I'll have a crash at launch.
Knowing how difficult it is to debug rare-occurrence segfaults, I wonder
if it is easier to just fix the above code and see if the launch crash
goes away for me.  If this were an issue, it'd be of the sort where
there is a race between the system setting up the color tables and the
GUI accessing them at launch.

Dan


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Dmitri A. Sergatskov
In reply to this post by John W. Eaton


On Thu, Feb 15, 2018 at 10:37 AM, John W. Eaton <[hidden email]> wrote:
I'd like to make the 4.2.2 release now.

One of the most visible issues it should fix is the crash on exit.  My build of the current stable sources does appear to work properly and teh crash no longer happens.  Does anyone else still experience the crash?

Thanks,

jwe



​I do not see the crash with

​LD_PRELOAD=/usr/lib64/libGLX_mesa.so.0 ./run-octave -q -f --no-gui
octave:1> __octave_config_info__ ("hg_id")
ans = 10e8dc9d453c

(still do with 4.2.1)

make check fails on __osmesa_print__.cc-tst with size mismatch
(I guess it was changed on dev version and not on stable?)


***** testif HAVE_OPENGL, HAVE_OSMESA, HAVE_GL2PS_H
 if (isunix ())
   hf = figure ("visible", "off");
   fn = tempname ();
   unwind_protect
     sombrero ();
     __osmesa_print__ (hf, fn, "svg");
     assert (stat (fn).size, 2579392, -0.1);
     img = __osmesa_print__ (hf);
     assert (size (img), [get(hf, "position")([4, 3]), 3]);
     ## Use pixel sum per RGB channel as fingerprint
     img_fp = squeeze (sum (sum (img), 2));
     assert (img_fp, [52942515; 54167797; 56158178], -0.05);
   unwind_protect_cleanup
     close (hf);
     unlink (fn);
   end_unwind_protect
 endif
!!!!! test failed
ASSERT errors for:  assert (stat (fn).size,2579392,-0.1)

  Location  |  Observed  |  Expected  |  Reason
     ()        3330663      2579392      Rel err 0.29126 exceeds tol 0.1

​This is on Fedora 27, gl2ps 1.4.0.

Dmitri.
--



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

John W. Eaton
Administrator
In reply to this post by Daniel Sebald
On 02/15/2018 12:52 PM, Daniel J Sebald wrote:

> On 02/15/2018 10:37 AM, John W. Eaton wrote:
>> I'd like to make the 4.2.2 release now.
>>
>> One of the most visible issues it should fix is the crash on exit.  My
>> build of the current stable sources does appear to work properly and
>> teh crash no longer happens.  Does anyone else still experience the
>> crash?
>>
>> Thanks,
>
> Not related to the exit crash, here's another one to consider before
> release:
>
> http://savannah.gnu.org/bugs/?53137
>
> I can't say this is a problem, and I can't point to any manifestation of
> the color table being accessed prematurely.  However, in doing
> development here, every once in a while I'll have a crash at launch.
> Knowing how difficult it is to debug rare-occurrence segfaults, I wonder
> if it is easier to just fix the above code and see if the launch crash
> goes away for me.  If this were an issue, it'd be of the sort where
> there is a race between the system setting up the color tables and the
> GUI accessing them at launch.

The variable editor doesn't exist on the stable branch, so it won't be
part of 4.2.2.

I pushed the patch to default.

jwe



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Andreas Weber-6
In reply to this post by Dmitri A. Sergatskov
Am 15.02.2018 um 19:32 schrieb Dmitri A. Sergatskov:

> make check fails on __osmesa_print__.cc-tst with size mismatch
> (I guess it was changed on dev version and not on stable?)
>
>
> ***** testif HAVE_OPENGL, HAVE_OSMESA, HAVE_GL2PS_H
>  if (isunix ())
>    hf = figure ("visible", "off");
>    fn = tempname ();
>    unwind_protect
>      sombrero ();
>      __osmesa_print__ (hf, fn, "svg");
>      assert (stat (fn).size, 2579392, -0.1);
>      img = __osmesa_print__ (hf);
>      assert (size (img), [get(hf, "position")([4, 3]), 3]);
>      ## Use pixel sum per RGB channel as fingerprint
>      img_fp = squeeze (sum (sum (img), 2));
>      assert (img_fp, [52942515; 54167797; 56158178], -0.05);
>    unwind_protect_cleanup
>      close (hf);
>      unlink (fn);
>    end_unwind_protect
>  endif
> !!!!! test failed
> ASSERT errors for:  assert (stat (fn).size,2579392,-0.1)
>
>   Location  |  Observed  |  Expected  |  Reason
>      ()        3330663      2579392      Rel err 0.29126 exceeds tol 0.1
>
> ​This is on Fedora 27, gl2ps 1.4.0.


Can you confirm, that you see a correct sombrero plot when running

hf = figure ("visible", "off");
sombrero ();
__osmesa_print__ (hf, "foo.svg", "svg");

Can you upload the resulting foo.svg and add a link?
Do you have a good idea how to check if the generated image is as
expected and which is more robust?

Thank you in advance, Andy


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Dmitri A. Sergatskov

__octave_config_info__ ("hg_id")
ans = 36ca11f78553

I think a binary format would be more robust.
(Stripping meta data and using high compression would probably improves things too)

Anyway, the svg produced by dev version (on the same computer) is slightly smaller
 ls -log sombrero_dev.svg
-rw-rw-r--. 1 2980162 Feb 16 01:27 sombrero_dev.svg
ls -log sombrero_stable.svg
-rw-rw-r--. 1 3330663 Feb 16 01:26 sombrero_stable.svg


On Fri, Feb 16, 2018 at 1:17 AM, Andreas Weber <[hidden email]> wrote:
Am 15.02.2018 um 19:32 schrieb Dmitri A. Sergatskov:
> make check fails on __osmesa_print__.cc-tst with size mismatch
> (I guess it was changed on dev version and not on stable?)
>
>
> ***** testif HAVE_OPENGL, HAVE_OSMESA, HAVE_GL2PS_H
>  if (isunix ())
>    hf = figure ("visible", "off");
>    fn = tempname ();
>    unwind_protect
>      sombrero ();
>      __osmesa_print__ (hf, fn, "svg");
>      assert (stat (fn).size, 2579392, -0.1);
>      img = __osmesa_print__ (hf);
>      assert (size (img), [get(hf, "position")([4, 3]), 3]);
>      ## Use pixel sum per RGB channel as fingerprint
>      img_fp = squeeze (sum (sum (img), 2));
>      assert (img_fp, [52942515; 54167797; 56158178], -0.05);
>    unwind_protect_cleanup
>      close (hf);
>      unlink (fn);
>    end_unwind_protect
>  endif
> !!!!! test failed
> ASSERT errors for:  assert (stat (fn).size,2579392,-0.1)
>
>   Location  |  Observed  |  Expected  |  Reason
>      ()        3330663      2579392      Rel err 0.29126 exceeds tol 0.1
>
> ​This is on Fedora 27, gl2ps 1.4.0.


Can you confirm, that you see a correct sombrero plot when running

hf = figure ("visible", "off");
sombrero ();
__osmesa_print__ (hf, "foo.svg", "svg");

Can you upload the resulting foo.svg and add a link?
Do you have a good idea how to check if the generated image is as
expected and which is more robust?

Thank you in advance, Andy


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

sombrero_stable.svg.bz2 (381K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Daniel Sebald
In reply to this post by John W. Eaton
On 02/15/2018 02:40 PM, John W. Eaton wrote:

> On 02/15/2018 12:52 PM, Daniel J Sebald wrote:
>> On 02/15/2018 10:37 AM, John W. Eaton wrote:
>>> I'd like to make the 4.2.2 release now.
>>>
>>> One of the most visible issues it should fix is the crash on exit.  
>>> My build of the current stable sources does appear to work properly
>>> and teh crash no longer happens.  Does anyone else still experience
>>> the crash?
>>>
>>> Thanks,
>>
>> Not related to the exit crash, here's another one to consider before
>> release:
>>
>> http://savannah.gnu.org/bugs/?53137
>>
>> I can't say this is a problem, and I can't point to any manifestation
>> of the color table being accessed prematurely.  However, in doing
>> development here, every once in a while I'll have a crash at launch.
>> Knowing how difficult it is to debug rare-occurrence segfaults, I
>> wonder if it is easier to just fix the above code and see if the
>> launch crash goes away for me.  If this were an issue, it'd be of the
>> sort where there is a race between the system setting up the color
>> tables and the GUI accessing them at launch.
>
> The variable editor doesn't exist on the stable branch, so it won't be
> part of 4.2.2.
>
> I pushed the patch to default.

OK, thanks.  I pulled the latest changes this afternoon and haven't seen
a crash at launch since.  Hopefully that did something.

Dan


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: 4.2.2 release

Sébastien Villemot-2
In reply to this post by John W. Eaton
On Thu, Feb 15, 2018 at 11:37:05AM -0500, John W. Eaton wrote:
> I'd like to make the 4.2.2 release now.

If time permits, please also review and merge my patch for bug #46571 (classdef
numel user method not consulted).

 https://savannah.gnu.org/bugs/?46571

Best,

--
⢀⣴⠾⠻⢶⣦⠀  Sébastien Villemot
⣾⠁⢠⠒⠀⣿⡁  Debian Developer
⢿⡄⠘⠷⠚⠋⠀  http://sebastien.villemot.name
⠈⠳⣄⠀⠀⠀⠀  http://www.debian.org


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

signature.asc (849 bytes) Download Attachment