[Squeak 4.0] Mac VM to use for 4.0?

 
Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?

--
Ron

 
>>>>> "Ronald" == Ronald Spengler <[hidden email]> writes:

Ronald> Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?

If that's a 64-bit image, it won't run on my machine. :)

--
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion

 
On Mon, Mar 08, 2010 at 07:48:44PM -0800, Randal L. Schwartz wrote:
> >>>>> "Ronald" == Ronald Spengler <[hidden email]> writes:
>
> Ronald> Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?
>
> If that's a 64-bit image, it won't run on my machine. :)

On the contrary, you most definitely *can* run a 64-bit image
on your 32-bit machine:

  http://squeakvm.org/squeak64/faq.html#4

IIUC, John's recent VMs can take advantage of 64-bit processing
on a Mac, as well as the more usual 32-bit processing mode.

Dave

 
Not the image, the VM; since this is the Mac VM linked on squeak.org.
I'm thinking it makes more sense to just ship the old VM, since it's
effectively an old image, runs without closures, and has been more
thoroughly tested on the old VM. It's possible that some folks will
want the license without wanting Trunk's changes, and have no use for
closures, etc.

On Mon, Mar 8, 2010 at 7:48 PM, Randal L. Schwartz
<[hidden email]> wrote:


--
Ron

 
>>>>> "David" == David T Lewis <[hidden email]> writes:

David> On Mon, Mar 08, 2010 at 07:48:44PM -0800, Randal L. Schwartz wrote:
>> >>>>> "Ronald" == Ronald Spengler <[hidden email]> writes:
>>
Ronald> Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?
>>
>> If that's a 64-bit image, it won't run on my machine. :)

David> On the contrary, you most definitely *can* run a 64-bit image
David> on your 32-bit machine:

David>   http://squeakvm.org/squeak64/faq.html#4

I misspoke when I said "64-bit image".

What I meant was a 64-bit program.  I think you need Snow Leopard
for that, and I'm only running Leopard.


--
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion

 
On Mon, 2010-03-08 at 19:27 -0800, Ronald Spengler wrote:
> Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?
>

Is John McIntosh available to ask? That is where I got the VMs I tested,
http://www.smalltalkconsulting.com/. (His VM page should get an award
for anonymity.)

In the last tests I ran on a PowerPC iMac 64 - 32 did not do as well as
the older VM. Note the warning about being for testing only.

My results:

Sunday, 28 February 2010, 3:17:45 am, squeak
Unit test results from Gary Dunn, [hidden email]

Model Name: iMac G5
Model Identifier: PowerMac8,1
Processor Name: PowerPC G5 (3.0)
Processor Speed: 1.8 GHz
Number Of CPUs: 1
L2 Cache (per CPU): 512 KB
Memory: 512 MB
Bus Speed: 600 MHz
Boot ROM Version: 5.2.2f4
OS X 10.5.8
Squeak3.11-9371-alpha.image

VM Squeak 64-32 5.4b2.app from ftp.smalltalkconsulting.com
"Only use this VM for testing, it lacks mac menu integration."
2578 run, 2560 passes, 6 expected failures, 10 failures, 1 errors, 1
unexpected passes
run time 16 min.

Failure List
AllocationTest>>#testOneGigAllocation
ClosureCompilerTest>>#testDebuggerTempAccess
ClosureCompilerTest>>#testInjectIntoDecompilations
ClosureCompilerTest>>#testInjectIntoDecompiledDebugs
DebuggerUnwindBug>>#testUnwindDebuggerWithStep
FloatTest>>#testNaNCompare
LocaleTest>>#testIsFontAvailable
MCPackageTest>>#testUnload
ScaledDecimalTest>>#testAsNumberWithSuperfluousDecimalPoint
SqNumberParserTest>>#testFloatGradualUnderflow
SqNumberParserTest>>#testFloatPrintString

Error List
SMDependencyTest>>#test2


VM Squeak 4.2.2beta1U.app
2578 run, 2555 passes, 7 expected failures, 15 failures, 1 errors, 0
unexpected passes
run time 15 min.

Failure List
BitBltTest>>#testAllAlphasRgbAdd
BitBltTest>>#testAllAlphasRgbMax
BitBltTest>>#testAllAlphasRgbMin
BitBltTest>>#testAllAlphasRgbMinInvert
BitBltTest>>#testAllAlphasRgbSub
ClosureCompilerTest>>#testDebuggerTempAccess
ClosureCompilerTest>>#testInjectIntoDecompilations
ClosureCompilerTest>>#testInjectIntoDecompiledDebugs
DebuggerUnwindBug>>#testUnwindDebuggerWithStep
FloatTest>>#testNaNCompare
LocaleTest>>#testIsFontAvailable
MCPackageTest>>#testUnload
ScaledDecimalTest>>#testAsNumberWithSuperfluousDecimalPoint
SqNumberParserTest>>#testFloatGradualUnderflow
SqNumberParserTest>>#testFloatPrintString

Error List
SMDependencyTest>>#test2
--
Gary Dunn, Honolulu
[hidden email]
http://openslate.net/
http://e9erust.blogspot.com/
Sent from Slate001

Ok, it's a test VM, not really for production usage, plus it has little support for plugins in 64bit mode.
It also doesn't run on 10.4 or earlier, unlike the 4.2.2 VM which should run on 10.2 or later.

Garry ran it in 32bit mode on a Powerpc. It *should* have behaved like the 4.2.2beta1U VM. But
he reported these tests worked with V5, but not V4.

On 2010-03-08, at 10:19 PM, Gary Dunn wrote:

> BitBltTest>>#testAllAlphasRgbAdd
> BitBltTest>>#testAllAlphasRgbMax
> BitBltTest>>#testAllAlphasRgbMin
> BitBltTest>>#testAllAlphasRgbMinInvert
> BitBltTest>>#testAllAlphasRgbSub



Obviously that is interesting...

What would be more telling is how it behaves if you run the tests on a MacIntel machine.

BTW it does run on PowerPC 32bit, or MacIntel in 32 or 64bit mode.
In theory if you do a get info, and check the (Open in 32bit mode) then it should enable use of all the
4.x era plugins.

--
===========================================================================
John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
===========================================================================

 
On 09.03.2010, at 04:27, Ronald Spengler wrote:
>
> Any reason I shouldn't ship "Squeak 64-32 5.4b2.app.zip" with Squeak 4.0?

Yes, it's experimental. Use the latest stable VM, 4.2.2 IIRC.

- Bert -

 


Den 09.03.2010 07:49, skrev John M McIntosh: Those tests are for the overflow bugs in BitBlt plugin which were fixed
in a later version than 4.2.2 was built from. (but included in 5.X)
IIRC, some time ago you mentioned maybe releasing a new 4.2.X as well,
with an updated version of the plugins :)

Cheers,
Henry

On 09.03.2010, at 12:54, Henrik Johansen wrote:

I for one would very much like if all the stable VMs had a 4 as major version number ...

- Bert -

 
On Tue, Mar 09, 2010 at 12:57:05PM +0100, Bert Freudenberg wrote:
>
> I for one would very much like if all the stable VMs had a 4 as
> major version number ...

If there are no objections, I will update the #versionString in VMMaker
to '4.0.0'. This will cause future unix VM builds to have labels such
as '4.0.0-1345'.

The down side it that it implies a major functional change to the VM,
when no such change has happened.

Objections?

Dave
 

 
On Thu, Mar 11, 2010 at 08:27:48AM -0500, David T. Lewis wrote:
I have updated the #versionString for VMMaker to '4.0.0' in VMMaker-dtl.161
on SqueakSource.

Dave