[Pharo-project] Fwd: [squeak-dev] macintosh cocoa 32/64bit & 64/64bit VM goes BETA
john needs beta testers....Do not forget I know that we are all terribly busy.
Begin forwarded message:
From: John M McIntosh <johnmci@smalltalkconsulting.com> Date: December 2, 2009 11:57:43 PM GMT+01:00 To: Squeak Virtual Machine Development Discussion <vm-dev@lists.squeakfoundation.org>, The general-purpose Squeak developers list <squeak-dev@lists.squeakfoundation.org>, "Pharo-project@lists.gforge.inria.fr Development" <pharo-project@lists.gforge.inria.fr> Subject: [squeak-dev] macintosh cocoa 32/64bit & 64/64bit VM goes BETA Reply-To: johnmci@smalltalkconsulting.com, The general-purpose Squeak developers list <squeak-dev@lists.squeakfoundation.org>
Ok, it's time for the community to do some testing.
ESUG provided me with some funding to take the existing iPhone VM source and extend that to build a new hosting platform using Cocoa on the macintosh and to enable the VM to run in 32 or 64 bit mode and work with a 32 or 64bit image.
I have placed two Alpha test VM s in the experimental folder sub-folder COCOA found in the usual places via
http://smalltalkconsulting.com/squeak.html
There are two VM's
(a) a 32/64 bit VM that runs 32bit squeak images in 32bit (powerpc, intel) or 64bit (intel) mode. (b) a 64/64 bit VM that runs 64bit squeak images in 64bit (intel).
The VM currently runs on 10.5.x and 10.6.x Support for 10.4.x might happen, we will see.
Until I sort out the re-integration of the source code into the squeakvm.org SVN tree you will find the source at isqueak.org
Issues, Oh lots. It's a fairly high mountain of material that needs to be tested.
Things that should work.
(a) access to files and directories UTF8 support only. (b) Full screen support (c) Sockets (d) Basic Sound (better than in the 3.x/4.x VMs) (e) jpegplugin2 (f) cut/copy/paste of text data (from/to) squeak and respect UTF8 chars (g) drag and drop of files into squeak
Right now it does not:
(a) Support a internet browser as a browser plugin. (b) The 64bit VM has no plugins at the moment. (b2) I've not included any plugins with the 32/64bit VM. Existing plugins might work if the VM is running as "Open in 32bit mode" someone can confirm that.
(c) No macintosh host menus, or serial port support (d) No Areithfa Ffenestri multiple-window support (e) No 1, 2, 4, 8, 16 bit appearance support.
Things I know about. (a) double clicking *.images when the 5.x VM is running sometimes does not open a new VM. (b) tossing a 32 image at the 64/64 bit VM or a 64 image at the 32/64 VM doesn't give visual feedback that it won't open the image. (c) Full screen grabs all the screens, yes that will change
Thoughts on internal plugins. Hint we should have TEST cases so we can confirm if all this stuff works. It would be extremely helpful if someone wanted to work on that. No doubt someone *could* run all the current tests for a 32bit image and see what breaks. However you'll need to explain where and what, versus Oh it just crashed...
ADPCMCodecPlugin, untested. Squeak3D, Likely busted due to use of (int) data types in 64bit mode. B2DPlugin, untested. BitBltPlugin, untested. BMPReadWriterPlugin, untested. ZipPlugin, untested. DSAPrims, untested. SqueakFFIPrims, untested. FFTPlugin, untested. FloatArrayPlugin, untested. GeniePlugin, untested. HostWindowPlugin, No support yet JPEGReaderPlugin, untested. Klatt, untested. LargeIntegers, untested. Matrix2x3Plugin, untested. MiscPrimitivePlugin, untested. ObjectiveCPlugin, untested. RePlugin, Likely busted due to use of (int) data types in 64bit mode. SecurityPlugin, untested. SoundCodecPrims, Likely busted due to use of (int) data types in 64bit mode. SoundGenerationPlugin, Likely busted due to use of (int) data types in 64bit mode. StarSqueakPlugin, untested. SurfacePlugin, untested. UUIDPlugin untested.
Possibly Large Integers may have a problem in 64bit image mode, but we have to have some discussion if the (unsigned int *) used by one routine is on purpose or an oversight.
64 bit images.
We need to built a 64bit Closure image. **** I've no time to do that. ***
David T Lewis provided a somewhat current Squeak.image which you'll find on the ftp site.
If you build a 64bit image you need to do that on a PowerPC, or make a copy of your VM on macIntel and flip the "Open in 32-bit mode" and the "Open in Rosetta" to on. You will only see the "Open in 32bit mode" if it's a 64bit VM, and the "Open in Rosetta on os-x for 10.5 or 10.6 (if you installed it) Follow the obvious instructions.
No doubt some is required to take the code and build a MC change set for Pharo or Squeak Thrunk use so that anyone can convert an existing image.
Also see http://bugs.squeak.org/view.php?id=5239 http://bugs.squeak.org/view.php?id=5240
Building your own VM?
Hopefully someone will attempt this so I can see if I checked in all the changes.
32bit image VM pick SqueakPureObjc and set #define SQ_VI_BYTES_PER_WORD 4 for a 64bit image reading VM pick SqueakPureObjc64*64 and set #define SQ_VI_BYTES_PER_WORD 8 in the required places.
PS You'll find that building a 32/64 VM using GNU's tool chain will give you lousy performance when running as a 64bit VM, more on this later.
---------------------------------------------- Oh and lastly I'm looking for some paid work (smalltalk or objective-c (mac/iPhone) If anyone requires a Smalltalk consultant, or an iPhone app built, please email me.
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter... -- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Hi, I did a quick test on the 32bit version on 10.5.8. I took a pharo 1.1 core 11049 and told it to update from the network. It processed all the updates (took some time) and drew the "ok" dialog. However it then hung. I am not sure if that is related to the VM or not. in a spare moment I will dig deeper. thanks, Mike On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Â Twitter: Â squeaker68882 Corporate Smalltalk Consulting Ltd. Â http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
That's pretty normal, it's happened to me a lot of times, at least. If you leave if for a while, it'll starts responding again. No idea why it happens though, I glanced at the code one time it happened, but no neurons fired :/ Cheers, Henry On 07.12.2009 21:49, Michael Roberts wrote:
Hi, I did a quick test on the 32bit version on 10.5.8. I took a pharo 1.1 core 11049 and told it to update from the network. It processed all the updates (took some time) and drew the "ok" dialog. However it then hung. I am not sure if that is related to the VM or not. in a spare moment I will dig deeper.
thanks, Mike
On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh<johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
ok thanks, I shall be more patient next time! Mike On Mon, Dec 7, 2009 at 9:33 PM, Henrik Sperre Johansen <henrik.s.johansen@veloxit.no> wrote:
That's pretty normal, it's happened to me a lot of times, at least. If you leave if for a while, it'll starts responding again. No idea why it happens though, I glanced at the code one time it happened, but no neurons fired :/
Cheers, Henry
On 07.12.2009 21:49, Michael Roberts wrote:
Hi, I did a quick test on the 32bit version on 10.5.8. Â I took a pharo 1.1 core 11049 and told it to update from the network. Â It processed all the updates (took some time) and drew the "ok" dialog. Â However it then hung. I am not sure if that is related to the VM or not. Â in a spare moment I will dig deeper.
thanks, Mike
On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> Â wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh<johnmci@smalltalkconsulting.com> Â Â Twitter: Â squeaker68882 Corporate Smalltalk Consulting Ltd. Â http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Ah, if someone can describe the test case: which I think is take a pharo 1.1 core 11049 then do update? I'll see where it's hung. would that be the pharo1.1-11073-alphadev09.12.1.zip or pharo1.1-11073-alphaweb09.12.1.zip from the 1.1 alpha links on the main pharo web page? On 2009-12-07, at 1:33 PM, Henrik Sperre Johansen wrote:
That's pretty normal, it's happened to me a lot of times, at least. If you leave if for a while, it'll starts responding again. No idea why it happens though, I glanced at the code one time it happened, but no neurons fired :/
Cheers, Henry
On 07.12.2009 21:49, Michael Roberts wrote:
Hi, I did a quick test on the 32bit version on 10.5.8. I took a pharo 1.1 core 11049 and told it to update from the network. It processed all the updates (took some time) and drew the "ok" dialog. However it then hung. I am not sure if that is related to the VM or not. in a spare moment I will dig deeper.
thanks, Mike
On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh<johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Hi John, it was not the pharo image, but the pharo core image. You can get them direct from the inria download area. This is linked off the pharo-project site. 11049 was an internal update so start from http://gforge.inria.fr/frs/download.php/24763/PharoCore-1.1-11043-alpha.zip I will do this myself just to check the test case. thanks, Mike On Monday, December 7, 2009, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
Ah, if someone can describe the test case:
which I think is take a pharo 1.1 core 11049 then do update? I'll see where it's hung.
would that be the pharo1.1-11073-alphadev09.12.1.zip or pharo1.1-11073-alphaweb09.12.1.zip
from the 1.1 alpha links on the main pharo web page?
On 2009-12-07, at 1:33 PM, Henrik Sperre Johansen wrote:
That's pretty normal, it's happened to me a lot of times, at least. If you leave if for a while, it'll starts responding again. No idea why it happens though, I glanced at the code one time it happened, but no neurons fired :/
Cheers, Henry
On 07.12.2009 21:49, Michael Roberts wrote:
Hi, I did a quick test on the 32bit version on 10.5.8. Â I took a pharo 1.1 core 11049 and told it to update from the network. Â It processed all the updates (took some time) and drew the "ok" dialog. Â However it then hung. I am not sure if that is related to the VM or not. Â in a spare moment I will dig deeper.
thanks, Mike
On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> Â wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh<johnmci@smalltalkconsulting.com> Â Â Twitter: Â squeaker68882 Corporate Smalltalk Consulting Ltd. Â http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Â Twitter: Â squeaker68882 Corporate Smalltalk Consulting Ltd. Â http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Hi, This "image need a while after update" is normal. I think it has to do with the really bad Weak* stuff... We should look at the fixes that Martin von Loewis did back in end of 2007 when he visited SCG. Marcus On Dec 8, 2009, at 8:46 PM, Michael Roberts wrote:
Hi John, it was not the pharo image, but the pharo core image. You can get them direct from the inria download area. This is linked off the pharo-project site. 11049 was an internal update so start from
http://gforge.inria.fr/frs/download.php/24763/PharoCore-1.1-11043-alpha.zip
I will do this myself just to check the test case.
thanks, Mike
On Monday, December 7, 2009, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
Ah, if someone can describe the test case:
which I think is take a pharo 1.1 core 11049 then do update? I'll see where it's hung.
would that be the pharo1.1-11073-alphadev09.12.1.zip or pharo1.1-11073-alphaweb09.12.1.zip
from the 1.1 alpha links on the main pharo web page?
On 2009-12-07, at 1:33 PM, Henrik Sperre Johansen wrote:
That's pretty normal, it's happened to me a lot of times, at least. If you leave if for a while, it'll starts responding again. No idea why it happens though, I glanced at the code one time it happened, but no neurons fired :/
Cheers, Henry
On 07.12.2009 21:49, Michael Roberts wrote:
Hi, I did a quick test on the 32bit version on 10.5.8. I took a pharo 1.1 core 11049 and told it to update from the network. It processed all the updates (took some time) and drew the "ok" dialog. However it then hung. I am not sure if that is related to the VM or not. in a spare moment I will dig deeper.
thanks, Mike
On Fri, Dec 4, 2009 at 7:33 PM, John M McIntosh <johnmci@smalltalkconsulting.com> wrote:
On 2009-12-04, at 4:30 AM, Stéphane Ducasse wrote:
john needs beta testers....Do not forget I know that we are all terribly busy.
Yes , *cough* feedback and response has been zero... Maybe I should push the self-destruct button on the 3.x, 4.x VM series. Where is that red button? Must be here somewhere in the office clutter...
-- =========================================================================== John M. McIntosh<johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
ok that's useful to know. it's a poor test case, i could not recreate it! would be good to add the fixes to the tracker. cheers, Mike On Wed, Dec 9, 2009 at 1:05 PM, Marcus Denker <denker@acm.org> wrote:
Hi,
This "image need a while after update" is normal.
participants (5)
-
Henrik Sperre Johansen -
John M McIntosh -
Marcus Denker -
Michael Roberts -
Stéphane Ducasse