Camera sig fault on 64 bits machines.

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
43 messages Options
123
Reply | Threaded
Open this post in threaded view
|

Camera sig fault on 64 bits machines.

Gustavo Duarte-2
 
Hi folks,

Imagen capture using web camera produces a a segmentation fault.

I'm using:

Squeak: 4.4.7
Scratch: 1.4.06
Plugins squeak-plugins-scratch: 1.4.0.2~svn.r83-1


My laptop is running Ubuntu 12.04 64 bits.

Here the output stack trace.
********************************************************************
gustavo@gusHP ~ $ scratch
Executing: padsp /usr/lib/squeak/4.4.7-2357/squeakvm -encoding UTF-8
-vm-display-x11 -fullscreen -plugins
/usr/lib/scratch/plugins/:/usr/lib/squeak/4.4.7-2357/ -vm-sound-oss
/usr/share/scratch/Scratch.image

Segmentation fault

11283804 BitBlt>setDestForm:
11283712 BitBlt class>toForm:
11283464 WarpBlt class>toForm:
11283372 ScratchCameraDialog>step
11283256 Morph>stepAt:
11280616 [] in PasteUpMorph>runStepMethods
11280120 OrderedCollection>do:
11280340 [] in PasteUpMorph>runStepMethods
11279452 BlockContext>ifError:
11279360 PasteUpMorph>runStepMethods
11271628 PasteUpMorph>doOneCycleNow
11271536 PasteUpMorph>doOneCycle
4967012 [] in Project>spawnNewProcess
4967104 [] in BlockContext>newProcess
Aborted
**********************************************


Where do you think could be the bug, on Squeak, Scratch or Plugin ?

Thanks in advance.

Gustavo.
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

David T. Lewis
 
On Sat, Apr 05, 2014 at 11:25:54AM -0300, Gustavo Duarte wrote:

>
> Hi folks,
>
> Imagen capture using web camera produces a a segmentation fault.
>
> I'm using:
>
> Squeak: 4.4.7
> Scratch: 1.4.06
> Plugins squeak-plugins-scratch: 1.4.0.2~svn.r83-1
>
>
> My laptop is running Ubuntu 12.04 64 bits.
>
> Here the output stack trace.
> ********************************************************************
> gustavo@gusHP ~ $ scratch
> Executing: padsp /usr/lib/squeak/4.4.7-2357/squeakvm -encoding UTF-8
> -vm-display-x11 -fullscreen -plugins
> /usr/lib/scratch/plugins/:/usr/lib/squeak/4.4.7-2357/ -vm-sound-oss
> /usr/share/scratch/Scratch.image
>
> Segmentation fault
>
> 11283804 BitBlt>setDestForm:
> 11283712 BitBlt class>toForm:
> 11283464 WarpBlt class>toForm:
> 11283372 ScratchCameraDialog>step
> 11283256 Morph>stepAt:
> 11280616 [] in PasteUpMorph>runStepMethods
> 11280120 OrderedCollection>do:
> 11280340 [] in PasteUpMorph>runStepMethods
> 11279452 BlockContext>ifError:
> 11279360 PasteUpMorph>runStepMethods
> 11271628 PasteUpMorph>doOneCycleNow
> 11271536 PasteUpMorph>doOneCycle
> 4967012 [] in Project>spawnNewProcess
> 4967104 [] in BlockContext>newProcess
> Aborted
> **********************************************
>
>
> Where do you think could be the bug, on Squeak, Scratch or Plugin ?

Hi Gustavo,

This is probably a bug in the camera plugin. This plugin has not been
updated to work properly in 64-bit mode, and it might well cause a crash
like this.

I think that you are probably using a VM that was compiled in 64-bit mode
for an Ubuntu distribution. A 64-bit VM works fine for most Squeak users
(I use one every day without problems). However, there are some plugins
that do not yet work in 64-bit mode, and this causes problems if those
plugins are used. For this reason, the VMs that are distributed from the
squeakvm.org or squeak.org sites are compiled in 32-bit mode.

If you can, please try installing a VM from http://squeakvm.org/unix/ and
using that with Scratch. Because it is a 32-bit application, you may also
need to install 32-bit compatibility libraries for Ubuntu. I think that
this will probably resolve the camera plugin problem.

I have also added a bug report on our Mantis bug tracker to follow up
on this issue at http://bugs.squeak.org/view.php?id=7816

Dave

Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

Gustavo Duarte-2
 
Hi Dave,

I followed your suggestion and could resolve this issue, thanks a lot.

Here the steps followed:

1) Uninstall, squeak-vm 64 bits comes with Ubuntu distribution, without
its depends.

1) dpkg -r --force-no-depends squeak-vm

2) Download squeak-vm 32 bits.

wget http://www.squeakvm.org/unix/release/Squeak-4.10.2.2614-linux_i386.sh

sudo sh Squeak-4.10.2.2614-linux_i386.sh

Choice  as path prefix /usr instead default: /usr/local


3) Install 32 bits compatibility library:

sudo apt-get install ia32-libs

I know that is a workaround, the finally solution as you said, is update
the plugins to working properly on 64 bit architectures.



Thank you very much !

Gustavo.




On 04/05/2014 12:44 PM, David T. Lewis wrote:

>
> On Sat, Apr 05, 2014 at 11:25:54AM -0300, Gustavo Duarte wrote:
>>
>> Hi folks,
>>
>> Imagen capture using web camera produces a a segmentation fault.
>>
>> I'm using:
>>
>> Squeak: 4.4.7
>> Scratch: 1.4.06
>> Plugins squeak-plugins-scratch: 1.4.0.2~svn.r83-1
>>
>>
>> My laptop is running Ubuntu 12.04 64 bits.
>>
>> Here the output stack trace.
>> ********************************************************************
>> gustavo@gusHP ~ $ scratch
>> Executing: padsp /usr/lib/squeak/4.4.7-2357/squeakvm -encoding UTF-8
>> -vm-display-x11 -fullscreen -plugins
>> /usr/lib/scratch/plugins/:/usr/lib/squeak/4.4.7-2357/ -vm-sound-oss
>> /usr/share/scratch/Scratch.image
>>
>> Segmentation fault
>>
>> 11283804 BitBlt>setDestForm:
>> 11283712 BitBlt class>toForm:
>> 11283464 WarpBlt class>toForm:
>> 11283372 ScratchCameraDialog>step
>> 11283256 Morph>stepAt:
>> 11280616 [] in PasteUpMorph>runStepMethods
>> 11280120 OrderedCollection>do:
>> 11280340 [] in PasteUpMorph>runStepMethods
>> 11279452 BlockContext>ifError:
>> 11279360 PasteUpMorph>runStepMethods
>> 11271628 PasteUpMorph>doOneCycleNow
>> 11271536 PasteUpMorph>doOneCycle
>> 4967012 [] in Project>spawnNewProcess
>> 4967104 [] in BlockContext>newProcess
>> Aborted
>> **********************************************
>>
>>
>> Where do you think could be the bug, on Squeak, Scratch or Plugin ?
>
> Hi Gustavo,
>
> This is probably a bug in the camera plugin. This plugin has not been
> updated to work properly in 64-bit mode, and it might well cause a crash
> like this.
>
> I think that you are probably using a VM that was compiled in 64-bit mode
> for an Ubuntu distribution. A 64-bit VM works fine for most Squeak users
> (I use one every day without problems). However, there are some plugins
> that do not yet work in 64-bit mode, and this causes problems if those
> plugins are used. For this reason, the VMs that are distributed from the
> squeakvm.org or squeak.org sites are compiled in 32-bit mode.
>
> If you can, please try installing a VM from http://squeakvm.org/unix/ and
> using that with Scratch. Because it is a 32-bit application, you may also
> need to install 32-bit compatibility libraries for Ubuntu. I think that
> this will probably resolve the camera plugin problem.
>
> I have also added a bug report on our Mantis bug tracker to follow up
> on this issue at http://bugs.squeak.org/view.php?id=7816
>
> Dave
>
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

David T. Lewis
 
On Sun, Apr 06, 2014 at 10:41:08AM -0300, Gustavo Duarte wrote:

>
> Hi Dave,
>
> I followed your suggestion and could resolve this issue, thanks a lot.
>
> Here the steps followed:
>
> 1) Uninstall, squeak-vm 64 bits comes with Ubuntu distribution, without
> its depends.
>
> 1) dpkg -r --force-no-depends squeak-vm
>
> 2) Download squeak-vm 32 bits.
>
> wget http://www.squeakvm.org/unix/release/Squeak-4.10.2.2614-linux_i386.sh
>
> sudo sh Squeak-4.10.2.2614-linux_i386.sh
>
> Choice  as path prefix /usr instead default: /usr/local
>
>
> 3) Install 32 bits compatibility library:
>
> sudo apt-get install ia32-libs
>
> I know that is a workaround, the finally solution as you said, is update
> the plugins to working properly on 64 bit architectures.
>

Hi Gustavo,

Thanks for the follow up. I added your information to the Mantis issue
in case other Ubuntu users need to work around this problem.

  http://bugs.squeak.org/view.php?id=7816

Dave
tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
Hi David.

I have added this thread to my "keepers" folder. I will try to get to it after the SqueakSSL.

cordially,

tty
tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
In reply to this post by Gustavo Duarte-2
 
Hi All,

Just to clarify, the CameraPlugin is for WebCams correct?




Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

David T. Lewis
 
>  Hi All,
>
>
>
> Just to clarify, the CameraPlugin is for WebCams correct?
>

Yes, that's right.

tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
Thx.

Also, for a heads up on the SqueakSSL, I am going to switch gears for a bit and then approach the problem from the Pharo side--if the pharo works out of the box on my 64 bit machine with 32 bit libs, then it will help
me in identifying the issue on the Squeak side. If not, then I will start digging deeper on what I have.

thx for your help.


tty.

---- On Thu, 10 Apr 2014 07:39:47 -0700 David T. Lewis<[hidden email]> wrote ----


> Hi All,
>
>
>
> Just to clarify, the CameraPlugin is for WebCams correct?
>

Yes, that's right.


Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

David T. Lewis
 
r.e. SqueakSSL: The plugin should currently be identical (I hope) in all
branches of the VM. Both the interpreter VM (trunk) and the Pharo VM use
CMake for the unix build, so you should not experience the libtool
difficulties when working with either of them. IIRC, Ian Piumarta
originally created the CMake build for Squeak VM back in 2009 largely
because of problems with libtool, particularly related mixed 32 and 64 bit
libraries. CMake seems to take care of the problem nicely.

>  Thx.
>
>
>
> Also, for a heads up on the SqueakSSL, I am going to switch gears for a
> bit and then approach the problem from the Pharo side--if the pharo works
> out of the box on my 64 bit machine with 32 bit libs, then it will help
> me in identifying the issue on the Squeak side. If not, then I will start
> digging deeper on what I have.
>
>
> thx for your help.
>
>
>
>
> tty.
>
> ---- On Thu, 10 Apr 2014 07:39:47 -0700 David T.
> Lewis&lt;[hidden email]&gt; wrote ----
>
>
>
> &gt; Hi All,
> &gt;
> &gt;
> &gt;
> &gt; Just to clarify, the CameraPlugin is for WebCams correct?
> &gt;
>
> Yes, that's right.
>
>
>
>


tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
---- On Thu, 10 Apr 2014 09:08:30 -0700 David T. Lewis <[hidden email]> wrote ----


>r.e. SqueakSSL: The plugin should currently be identical (I hope) in all 
>branches of the VM. Both the interpreter VM (trunk) and the Pharo VM use 
>CMake for the unix build, so you should not experience the libtool 
>difficulties when working with either of them. 
>IIRC, Ian Piumarta  originally created the CMake build for Squeak VM back in 2009 largely 
>because of problems with libtool, particularly related mixed 32 and 64 bit 
>libraries. CMake seems to take care of the problem nicely.

Thx David


A couple of questions...

1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build tree from CMake (that is what I have inferred from emails he has addressed to Tim R)?

2. If it is going to be Gnu-ified (which I do prefer) would separate build trees for 64 bit, 32 bit and 64/32 hybrid build trees make sense for simplifying and isolating problematic build issues?
      (If I may add, for the SqueakSSL, I reduced the source to the minimum to get the unixbuild to compile and run--AsynchFilePlugin \ B2DPlugin \ BitBltPlugin \FilePlugin \SocketPlugin \MiscPrimitivePlugin
       and a very slimmed down source tree that is actually easy to understand--see my blog post: http://timmydosmalltalk.wordpress.com/2014/04/08/first-take-on-the-minimum-plugins-needed-for-a-functioning-image/)


Regardless,we will git-r-done. Its going to take me many more brain cycles to get good at this stuff, so it will take time. 









Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

timrowledge


On 10-04-2014, at 9:51 AM, gettimothy <[hidden email]> wrote:
>
> 1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build tree from CMake (that is what I have inferred from emails he has addressed to Tim R)?

Eliot has stated that he would be happy to see the cog stuff made properly compatible with everything else and moved to using CMake - but that he has no interest in doing nor time to do the work. I offered a $1000 bounty for anyone (or several) making Cog work with CMake, but nobody appears to have any need for a spare $1000 so far.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Religious tolerance


tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
As it just so happens, I am unemployed and broke!


Please define the project scope and deliverables and I will get it done!




---- On Thu, 10 Apr 2014 10:07:58 -0700 tim Rowledge<[hidden email]> wrote ----



On 10-04-2014, at 9:51 AM, gettimothy <[hidden email]> wrote:
>
> 1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build tree from CMake (that is what I have inferred from emails he has addressed to Tim R)?

Eliot has stated that he would be happy to see the cog stuff made properly compatible with everything else and moved to using CMake - but that he has no interest in doing nor time to do the work. I offered a $1000 bounty for anyone (or several) making Cog work with CMake, but nobody appears to have any need for a spare $1000 so far.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Religious tolerance



Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

David T. Lewis
In reply to this post by tty
 
>
> A couple of questions...
>
>
> 1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build
> tree from CMake (that is what I have inferred from emails he has addressed
> to Tim R)?
>

See Tim's answer. I would love to see someone get the $1000 prize :-)

>
> 2. If it is going to be Gnu-ified (which I do prefer) would separate build
> trees for 64 bit, 32 bit and 64/32 hybrid build trees make sense for
> simplifying and isolating problematic build issues?

Yes, you will want to use different build directories for this.

>       (If I may add, for the SqueakSSL, I reduced the source to the
> minimum to get the unixbuild to compile and run--AsynchFilePlugin \
> B2DPlugin \ BitBltPlugin \FilePlugin \SocketPlugin
> \MiscPrimitivePlugin
>        and a very slimmed down source tree that is actually easy to
> understand--see my blog post:
> http://timmydosmalltalk.wordpress.com/2014/04/08/first-take-on-the-minimum-plugins-needed-for-a-functioning-image/)
>

If you are working in VMM trunk, try this:

    VMMakerTool minimal

Also, the VMMakerTool makes it easy to regenerate just a specific plugin,
after which you can just run make again in your build directory.

Dave


tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
David.

I am taking Tim up on his bounty offer. Gotta love a chance to contribute, learn something new and interesting and pay the rent.


I want to get get my webcam working today and I will turn my attention to CMake tomorrow. 


cheers

tty.

---- On Thu, 10 Apr 2014 11:14:00 -0700 David T. Lewis<[hidden email]> wrote ----


>
> A couple of questions...
>
>
> 1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build
> tree from CMake (that is what I have inferred from emails he has addressed
> to Tim R)?
>

See Tim's answer. I would love to see someone get the $1000 prize :-)

>
> 2. If it is going to be Gnu-ified (which I do prefer) would separate build
> trees for 64 bit, 32 bit and 64/32 hybrid build trees make sense for
> simplifying and isolating problematic build issues?

Yes, you will want to use different build directories for this.

> (If I may add, for the SqueakSSL, I reduced the source to the
> minimum to get the unixbuild to compile and run--AsynchFilePlugin \
> B2DPlugin \ BitBltPlugin \FilePlugin \SocketPlugin
> \MiscPrimitivePlugin
> and a very slimmed down source tree that is actually easy to
> understand--see my blog post:
> http://timmydosmalltalk.wordpress.com/2014/04/08/first-take-on-the-minimum-plugins-needed-for-a-functioning-image/)
>

If you are working in VMM trunk, try this:

VMMakerTool minimal

Also, the VMMakerTool makes it easy to regenerate just a specific plugin,
after which you can just run make again in your build directory.

Dave



Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

Eliot Miranda-2
In reply to this post by tty
 
Hi Tty,


On Thu, Apr 10, 2014 at 10:26 AM, gettimothy <[hidden email]> wrote:
 
As it just so happens, I am unemployed and broke!


Please define the project scope and deliverables and I will get it done!

I'm not paying so I don't get to state the deliverables.  But what I want is for the Cog linux build to use CMake.  That is, make the build directories under http://www.squeakvm.org/svn/squeak/branches/Cog/unixbuild and http://www.squeakvm.org/svn/squeak/branches/Cog/nscogbuild/unixbuild to work, but to use CMake.  My current build enbvironment is a CentOS 5.3 VM running under Parallels.  The most convenient thing for me would be that the builds continue to work on this now rather old CentOS.
 
Thanks!!





---- On Thu, 10 Apr 2014 10:07:58 -0700 tim Rowledge<[hidden email]> wrote ----



On 10-04-2014, at 9:51 AM, gettimothy <[hidden email]> wrote:
>
> 1. Regarding CMake Isn't Eliot migrating more towards a Gnu style build tree from CMake (that is what I have inferred from emails he has addressed to Tim R)?

Eliot has stated that he would be happy to see the cog stuff made properly compatible with everything else and moved to using CMake - but that he has no interest in doing nor time to do the work. I offered a $1000 bounty for anyone (or several) making Cog work with CMake, but nobody appears to have any need for a spare $1000 so far.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Religious tolerance







--
best,
Eliot
tty
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

tty
 
Hi Eliot.

>> But what I want is for the Cog linux build to use CMake.  That is, make the build directories under http://www.squeakvm.org/svn/squeak/branches/Cog/unixbuild and http://www.squeakvm.org/svn/squeak/branches/Cog/nscogbuild/unixbuild to work, but to use CMake.  My current >>build enbvironment is a CentOS 5.3 VM running under Parallels.  The most convenient thing for me would be that the builds continue to work on this now rather old CentOS.
 

I am not familiar with Parallels, but a quick search showed its some sort of virtualization product. Centos is linux for paranoid sys-admins, so I should be good to go there (I have a spare partition where my windows7  blew itself up, so I can install it there if need be)

I have a static ip so I can make my work available for review and comment as it progresses. 

cordially,

tty



Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

Eliot Miranda-2
 



On Thu, Apr 10, 2014 at 11:45 AM, gettimothy <[hidden email]> wrote:
 
Hi Eliot.

>> But what I want is for the Cog linux build to use CMake.  That is, make the build directories under http://www.squeakvm.org/svn/squeak/branches/Cog/unixbuild and http://www.squeakvm.org/svn/squeak/branches/Cog/nscogbuild/unixbuild to work, but to use CMake.  My current >>build enbvironment is a CentOS 5.3 VM running under Parallels.  The most convenient thing for me would be that the builds continue to work on this now rather old CentOS.
 

I am not familiar with Parallels, but a quick search showed its some sort of virtualization product. Centos is linux for paranoid sys-admins, so I should be good to go there (I have a spare partition where my windows7  blew itself up, so I can install it there if need be)

I have a static ip so I can make my work available for review and comment as it progresses. 

sounds good!
 

cordially,

tty







--
best,
Eliot
Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

timrowledge
In reply to this post by tty


On 10-04-2014, at 11:17 AM, gettimothy <[hidden email]> wrote:

> David.
>
> I am taking Tim up on his bounty offer. Gotta love a chance to contribute, learn something new and interesting and pay the rent.

OK. I’ll be very pleased if something can actually come of this. I’ve had several people claim they were going to solve the problem Real Soon Now but so far nothing. Forgive me if I seem world-weary and cynical but that’s because… I am.

The aim is to get Cog on *nix being built via a mechanism as near as possible to that used for the plain interp in the squeakvm.org trunk tree. Windows & Mac may or may not ever be targets and I don’t care right now.

If you read the archives of the main squeak list and the vm-dev list after searching for all messages with ‘cmake’ in the subject and dating back to last june you’ll find at least 60-some hits. After a quick scan of my local archive I see a variety of suggestions to think about. I note that Ron Teitelbaum wrote that Goran was working on the issue at 3DICC, for example.

The grand aim would be to unify things so well that Eliot throws off the chains of keeping a virtual fork and starts using the main trunk since it is so easy to do. I suspect that might be more work than can be done in a few weeks even. What *I* would be happy with would be a checkout that works on my Pi, produces a working stackvm, matches the trunk cmake setup as closely as possible and is clearly documented so it can be further developed towards the Grand Aim.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: RSC: Rewind System Clock


tty
Reply | Threaded
Open this post in threaded view
|

VM CMake Project

tty
 
Tim,

Thank you. 

I have changed the subject line to reflect the project. 


I will start reviewing the archives tomorrow and come up with a roadmap. I don't have any projects besides this right now so I can give it my full attention and hopefully get something intuitive, functional and documented in place by end of next week-ish.

Cordially,

tty.

---- On Thu, 10 Apr 2014 12:53:28 -0700 tim Rowledge<[hidden email]> wrote ----



On 10-04-2014, at 11:17 AM, gettimothy <[hidden email]> wrote:

> David.
>
> I am taking Tim up on his bounty offer. Gotta love a chance to contribute, learn something new and interesting and pay the rent.

OK. I’ll be very pleased if something can actually come of this. I’ve had several people claim they were going to solve the problem Real Soon Now but so far nothing. Forgive me if I seem world-weary and cynical but that’s because… I am.

The aim is to get Cog on *nix being built via a mechanism as near as possible to that used for the plain interp in the squeakvm.org trunk tree. Windows & Mac may or may not ever be targets and I don’t care right now.

If you read the archives of the main squeak list and the vm-dev list after searching for all messages with ‘cmake’ in the subject and dating back to last june you’ll find at least 60-some hits. After a quick scan of my local archive I see a variety of suggestions to think about. I note that Ron Teitelbaum wrote that Goran was working on the issue at 3DICC, for example.

The grand aim would be to unify things so well that Eliot throws off the chains of keeping a virtual fork and starts using the main trunk since it is so easy to do. I suspect that might be more work than can be done in a few weeks even. What *I* would be happy with would be a checkout that works on my Pi, produces a working stackvm, matches the trunk cmake setup as closely as possible and is clearly documented so it can be further developed towards the Grand Aim.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: RSC: Rewind System Clock



Reply | Threaded
Open this post in threaded view
|

Re: Camera sig fault on 64 bits machines.

Andreas Wacknitz
In reply to this post by timrowledge

While we are at it: Linux is not a UNIX. Even if we are putting it under the UNIX umbrella, it is not the only UNIX.

I am trying to get Squeak/Pharo to run under Solaris (OpenSolaris) with some trouble. But also BSD folks have difficulties because
UNIX == Linux in the view of the Squeak VM.
While my first try with recent SqueakVM sources wasn’t successful (I got "DosFileDirectory(Object)>>doesNotUnderstand: #pathFromURI:“ from the provided image), I was able to build a recent pharovm after I manipulated the image in order to get the vm-sound-Sun module instead of the hard coded vm-sound-ALSA for the UNIX platform. I got everything compiled but the resulting binary stuck after showing the complete graphics
with no reactions to mouse or keyboard events. I found out that inside the aioPoll function the call to select() always returns 0.
Because of my limited spare time I didn’t get further yet.

So, please consider not only the 3 major platforms (Windows, Mac OS, Linux) when doing the great work on the VM.

Best regards
Andreas
 
Am 10.04.2014 um 21:53 schrieb tim Rowledge <[hidden email]>:

>
>
> On 10-04-2014, at 11:17 AM, gettimothy <[hidden email]> wrote:
>
>> David.
>>
>> I am taking Tim up on his bounty offer. Gotta love a chance to contribute, learn something new and interesting and pay the rent.
>
> OK. I’ll be very pleased if something can actually come of this. I’ve had several people claim they were going to solve the problem Real Soon Now but so far nothing. Forgive me if I seem world-weary and cynical but that’s because… I am.
>
> The aim is to get Cog on *nix being built via a mechanism as near as possible to that used for the plain interp in the squeakvm.org trunk tree. Windows & Mac may or may not ever be targets and I don’t care right now.
>
> If you read the archives of the main squeak list and the vm-dev list after searching for all messages with ‘cmake’ in the subject and dating back to last june you’ll find at least 60-some hits. After a quick scan of my local archive I see a variety of suggestions to think about. I note that Ron Teitelbaum wrote that Goran was working on the issue at 3DICC, for example.
>
> The grand aim would be to unify things so well that Eliot throws off the chains of keeping a virtual fork and starts using the main trunk since it is so easy to do. I suspect that might be more work than can be done in a few weeks even. What *I* would be happy with would be a checkout that works on my Pi, produces a working stackvm, matches the trunk cmake setup as closely as possible and is clearly documented so it can be further developed towards the Grand Aim.
>
> tim
> --
> tim Rowledge; [hidden email]; http://www.rowledge.org/tim
> Strange OpCodes: RSC: Rewind System Clock
>
>

123