Windows 8

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

Windows 8

John W. Eaton
Administrator
Has anyone tested the Windows binary distribution of 4.0.0-rc2 on a
Windows 8 system?  I remember that there were reports of problems with
earlier versions and I'm wondering if those have been fixed.  If not,
what were/are the problems?

jwe

Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

giorgio.denunzio
Il 01/04/2015 18:19, John W. Eaton ha scritto:
> Has anyone tested the Windows binary distribution of 4.0.0-rc2 on a
> Windows 8 system?  I remember that there were reports of problems with
> earlier versions and I'm wondering if those have been fixed. If not,
> what were/are the problems?
>
> jwe
>
>

Hi John,
I have started to test rc2 on w7 and w8.
On w8, apart from the same problems I am having on w7, i.e.:
1) in the GUI, ctrl+C for script interrupting is masked by ctrl+C for
copy, so scripts are not interruptible; there is a workaround by
shortcut redefinition;
2) pause() does not work as expected (quite severe bug imho),
I am also experiencing a problem concerning figure saving to disk. The
problem is not present on all the W8 pcs on which I have tried, only on
2 over 3 of them.
When I plot something, eg a trivial x=1:10;plot(x,x.^2); the plot is ok.
When I save the image to a pdf file (which by the way and unfortunately
is the only available format), only the axes appear in the created pdf
file, no curve. This is really annoying, because at present on those pcs
the only way to take a plot and put it somewhere (e.g. a .doc file) is
by screen snapshots.
Best regards
Giorgio

--
Dr. Giorgio De Nunzio
Dipart. di Matematica e Fisica "Ennio De Giorgi", Univ. del Salento
INFN Sezione di Lecce
DReAM (Laboratorio Diffuso di Ricerca Interdisciplinare Applicata alla Medicina)
ADAM (Advanced Data Analysis in Medicine) srl
Lecce (Italy)
[hidden email]
[hidden email]
tel +39 0832 297084-297051
fax +39 0832 297100
mobile +39 320 3829845


---
Questa e-mail è stata controllata per individuare virus con Avast antivirus.
http://www.avast.com


Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

Søren Hauberg
In reply to this post by John W. Eaton
Den 01-04-2015 kl. 18:19 skrev John W. Eaton:
> Has anyone tested the Windows binary distribution of 4.0.0-rc2 on a
> Windows 8 system?  I remember that there were reports of problems with
> earlier versions and I'm wondering if those have been fixed. If not,
> what were/are the problems?
I recently got a new laptop, which comes with Windows 8 pre-installed. I
haven't had the time to wipe the machine clean and install Linux yet, so
I figured I could do a quick test of how things work on a fresh Windows 8.

I downloaded the rc2 .exe installer and ran it. I was greeted with a
warning saying "Do you want to allow the following program from an
unknown publisher to make changes to this computer?". I said yes. Would
it be possible to provide a publisher (e.g. GNU) ? That would make this
first step less scary.

I was then presented with a warning from the Octave installer saying
that Windows 8 is not supported -- do I really want to continue? I went
ahead. The installer was easy to use, so no troubles here.

I started Octave and ran the following commands,

   A = rand(7);
   C = chol(A * A');

Worked nicely. I then tried plotting

   plot(diag(C))

This opened a figure window at the very top of the screen; in fact the
top window decoration is outside the screen, so I can't actually grab
the window and move it around. Also, the plot took around one minute to
appear, which made me think that it wasn't working.

Other than that, I get the impression that things are working fairly
nicely in Windows 8. One minor issue is that I have a high resolution
screen (3800x1800, I think). While the main GUI is nicely scaled, the
OpenGL plots appear rather tiny -- perhaps these could be scaled?

Is there anything in particular you would like to see tested?

Cheers
Søren

Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

giorgio.denunzio
Il 01/04/2015 20:52, Søren Hauberg ha scritto:

> Den 01-04-2015 kl. 18:19 skrev John W. Eaton:
>> Has anyone tested the Windows binary distribution of 4.0.0-rc2 on a
>> Windows 8 system?  I remember that there were reports of problems
>> with earlier versions and I'm wondering if those have been fixed. If
>> not, what were/are the problems?
> I recently got a new laptop, which comes with Windows 8 pre-installed.
> I haven't had the time to wipe the machine clean and install Linux
> yet, so I figured I could do a quick test of how things work on a
> fresh Windows 8.
>
> I downloaded the rc2 .exe installer and ran it. I was greeted with a
> warning saying "Do you want to allow the following program from an
> unknown publisher to make changes to this computer?". I said yes.
> Would it be possible to provide a publisher (e.g. GNU) ? That would
> make this first step less scary.
>
> I was then presented with a warning from the Octave installer saying
> that Windows 8 is not supported -- do I really want to continue? I
> went ahead. The installer was easy to use, so no troubles here.
>
> I started Octave and ran the following commands,
>
>   A = rand(7);
>   C = chol(A * A');
>
> Worked nicely. I then tried plotting
>
>   plot(diag(C))
>
> This opened a figure window at the very top of the screen; in fact the
> top window decoration is outside the screen, so I can't actually grab
> the window and move it around. Also, the plot took around one minute
> to appear, which made me think that it wasn't working.
>
> Other than that, I get the impression that things are working fairly
> nicely in Windows 8. One minor issue is that I have a high resolution
> screen (3800x1800, I think). While the main GUI is nicely scaled, the
> OpenGL plots appear rather tiny -- perhaps these could be scaled?
>
> Is there anything in particular you would like to see tested?
>
> Cheers
> Søren
>
>

Hi Søren,
thanks for your tests.
I think the delay in plotting only takes place the first time (at least
it was so till now for me on many win7/win8 pcs). The following plots
are quick. I imagine it depends on some sort of first-time house keeping.
On the contrary, I have never seen this "partially out of screen" issue.

Can you please save the figure as a pdf file, then open it to see if the
plot is well drawn?
I tried with the "sombrero" surface and with the simple x=1:10; plot(x,
x.^2) stuff, obtaining empty axes as the pdf file (on some installations).

Thanks
Giorgio



--
Dr. Giorgio De Nunzio
Dipart. di Matematica e Fisica "Ennio De Giorgi", Univ. del Salento
INFN Sezione di Lecce
DReAM (Laboratorio Diffuso di Ricerca Interdisciplinare Applicata alla Medicina)
ADAM (Advanced Data Analysis in Medicine) srl
Lecce (Italy)
[hidden email]
[hidden email]
tel +39 0832 297084-297051
fax +39 0832 297100
mobile +39 320 3829845


---
Questa e-mail è stata controllata per individuare virus con Avast antivirus.
http://www.avast.com


Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

Søren Hauberg


Den 01-04-2015 kl. 21:06 skrev Giorgio De Nunzio:
> Can you please save the figure as a pdf file, then open it to see if
> the plot is well drawn?
> I tried with the "sombrero" surface and with the simple x=1:10;
> plot(x, x.^2) stuff, obtaining empty axes as the pdf file (on some
> installations).
If I do

   sombrero
   print test.pdf
   print test.png

then I get two files (test.pdf, test.png) which both contain the axis
but not the sombrero. So plotting doesn't quite work.

I then tried the same with

   plot(1:10)

with the same result (axis is shown, but plot content is not).

If I do

   imshow(rand(100))
   print test.pdf
   print test.png

then I get files with the same content as is shown on screen.

I also tried

   image(255*rand(100))
   print test.pdf
   print test.png

and I get files that contain both the image as well as the axis.

So, sometimes plotting works; other times it doesn't.

Søren


>
> Thanks
> Giorgio
>
>
>


Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

John Frain
I ran all of the graphics tests on two PCs with Windows 8.1. Both
produced correct graphs on screen, or as pdf or png files. The first
Pc is an elderly Dell with an NVIDIA graphics card and 5 GB RAM.  The
second is a cheap wimdows 8.1 tablet with 1 GB RAM.

I can also report that I have run some programs on both with no problems.

The time to the first graph on the tablet is much faster (almost
instantaneous) on the tablet.  As noted previously this is very  slow
on any other PC that I have used. (SSD drive compared to hard disk??).
It is remarkable how well octave runs on such a low resource tablet.
Given the current tablet interface I would much prefer my desktop to
do serious work.

John


John C Frain, Ph.D.

Economics Department             3 Aranleigh Park
Trinity College Dublin                 Rathfarnham
College Green                           Dublin 14
Dublin 2                                    Ireland
Ireland
www.tcd.ie/Economics/staff/frainj/home.htm
mailto:[hidden email]
mailto:[hidden email]


On 1 April 2015 at 20:19, Søren Hauberg <[hidden email]> wrote:

>
>
> Den 01-04-2015 kl. 21:06 skrev Giorgio De Nunzio:
>>
>> Can you please save the figure as a pdf file, then open it to see if the
>> plot is well drawn?
>> I tried with the "sombrero" surface and with the simple x=1:10; plot(x,
>> x.^2) stuff, obtaining empty axes as the pdf file (on some installations).
>
> If I do
>
>   sombrero
>   print test.pdf
>   print test.png
>
> then I get two files (test.pdf, test.png) which both contain the axis but
> not the sombrero. So plotting doesn't quite work.
>
> I then tried the same with
>
>   plot(1:10)
>
> with the same result (axis is shown, but plot content is not).
>
> If I do
>
>   imshow(rand(100))
>   print test.pdf
>   print test.png
>
> then I get files with the same content as is shown on screen.
>
> I also tried
>
>   image(255*rand(100))
>   print test.pdf
>   print test.png
>
> and I get files that contain both the image as well as the axis.
>
> So, sometimes plotting works; other times it doesn't.
>
> Søren
>
>
>>
>> Thanks
>> Giorgio
>>
>>
>>
>
>

Reply | Threaded
Open this post in threaded view
|

Re: Re: Windows 8

Pantxo
In reply to this post by Søren Hauberg
Hi,

> Other than that, I get the impression that things are working fairly
> nicely in Windows 8. One minor issue is that I have a high resolution
> screen (3800x1800, I think). While the main GUI is nicely scaled, the
> OpenGL plots appear rather tiny -- perhaps these could be scaled?
>
>
What does the following script return on that machine and does the
computed size conform to the actual size of the screen?

dpi = get (0, "screenpixelsperinch")
sz = get (0, "screensize")
sprintf ("width: %f in, height: %f in\n", sz(3)/dpi, sz(4)/dpi)

> Søren
>
>

Thanks,

Pantxo

Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

Søren Hauberg


Den 02-04-2015 kl. 14:17 skrev Pantxo Diribarne:

> Hi,
>
>> Other than that, I get the impression that things are working fairly
>> nicely in Windows 8. One minor issue is that I have a high resolution
>> screen (3800x1800, I think). While the main GUI is nicely scaled, the
>> OpenGL plots appear rather tiny -- perhaps these could be scaled?
>>
>>
> What does the following script return on that machine and does the
> computed size conform to the actual size of the screen?
>
> dpi = get (0, "screenpixelsperinch")
> sz = get (0, "screensize")
> sprintf ("width: %f in, height: %f in\n", sz(3)/dpi, sz(4)/dpi)
I get

   dpi = 110.90
   sz = [1, 1, 1280, 720]
   width: 11.541991 in, height: 6.492370

This is a 13 inch screen (diagonal measure, I guess) running 3800x1800,
so these numbers don't appear to be quite right. I don't know if this is
due to the way Windows rescales its GUI in hidpi screeens or if it's due
to a bug in Octave.

Søren

Reply | Threaded
Open this post in threaded view
|

Re: Windows 8

Pantxo
Le 02/04/2015 15:08, Søren Hauberg a écrit :

>
>
>
>
> Den 02-04-2015 kl. 14:17 skrev Pantxo Diribarne:
>
>> Hi,
>>
>>
>>
>>> Other than that, I get the impression that things are working fairly
>>> nicely in Windows 8. One minor issue is that I have a high resolution
>>> screen (3800x1800, I think). While the main GUI is nicely scaled, the
>>> OpenGL plots appear rather tiny -- perhaps these could be scaled?
>>>
>>>
>>>
>>>
>>>
>> What does the following script return on that machine and does the
>> computed size conform to the actual size of the screen?
>>
>>
>>
>> dpi = get (0, "screenpixelsperinch")
>>
>> sz = get (0, "screensize")
>>
>> sprintf ("width: %f in, height: %f in\n", sz(3)/dpi, sz(4)/dpi)
>>
> I get
>
>
>
>   dpi = 110.90
>
>   sz = [1, 1, 1280, 720]
>
>   width: 11.541991 in, height: 6.492370
>
>
>
> This is a 13 inch screen (diagonal measure, I guess) running 3800x1800,
> so these numbers don't appear to be quite right. I don't know if this is
> due to the way Windows rescales its GUI in hidpi screeens or if it's due
> to a bug in Octave.
>
>
>
> Søren
>
>
>
Can you file a bug report anyway so that this does not get lost?
Thanks.

Pantxo