Smalltalk › Squeak › Squeak - Dev

#'insufficient object memory' in Spur

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

4 messages Options

Chris Muller-4

#'insufficient object memory' in Spur

Hi Eliot, thanks a lot for your help on the CM header. I decided to
try to add a new bit indicator in my CompiledMethodBuffer (pickled
state of a CM) to indicate which header format it was pickled in so I
can hopefully maintain a heterogeneous environment.

Now, maybe I goofed up my image while testing it at some point because
I cannot seem to allocate a 50MB ByteString (even though (Smalltalk
vmParameterAt: 3) reports just 89,627,904)

String new: 50000000

causes a failure in primitve 71, which causes the failure code to run,
which sends #growMemoryByAtLeast: to Smalltalk. But, there are no
implementors of #growMemoryByAtLeast: in the image..!

Do you have an impl. of that method? This could be a real problem for
me to move to Spur because I absolutely depend on handling OutOfMemory
errors to build my large models..

Eliot Miranda-2

Re: #'insufficient object memory' in Spur

Hi Chris,

On Fri, Nov 14, 2014 at 7:43 AM, Chris Muller <[hidden email]> wrote:

Hi Eliot, thanks a lot for your help on the CM header. I decided to
try to add a new bit indicator in my CompiledMethodBuffer (pickled
state of a CM) to indicate which header format it was pickled in so I
can hopefully maintain a heterogeneous environment.

Now, maybe I goofed up my image while testing it at some point because
I cannot seem to allocate a 50MB ByteString (even though (Smalltalk
vmParameterAt: 3) reports just 89,627,904)

String new: 50000000

causes a failure in primitve 71, which causes the failure code to run,
which sends #growMemoryByAtLeast: to Smalltalk. But, there are no
implementors of #growMemoryByAtLeast: in the image..!

If you download a recent Spur image from my web site you'll find it there-in.

Do you have an impl. of that method? This could be a real problem for
me to move to Spur because I absolutely depend on handling OutOfMemory
errors to build my large models..

best,

Eliot

Chris Muller-4

Re: #'insufficient object memory' in Spur

I was already using the 10/28 image. Ah ha! It seems I accidently
Load'ed that latest System package (unconverted) instead of Merge'ing
it.. OOPS! Not my day today!

On Fri, Nov 14, 2014 at 2:26 PM, Eliot Miranda <[hidden email]> wrote:

> Hi Chris,
>
> On Fri, Nov 14, 2014 at 7:43 AM, Chris Muller <[hidden email]> wrote:
>>
>> Hi Eliot, thanks a lot for your help on the CM header. I decided to
>> try to add a new bit indicator in my CompiledMethodBuffer (pickled
>> state of a CM) to indicate which header format it was pickled in so I
>> can hopefully maintain a heterogeneous environment.
>>
>> Now, maybe I goofed up my image while testing it at some point because
>> I cannot seem to allocate a 50MB ByteString (even though (Smalltalk
>> vmParameterAt: 3) reports just 89,627,904)
>>
>> String new: 50000000
>>
>> causes a failure in primitve 71, which causes the failure code to run,
>> which sends #growMemoryByAtLeast: to Smalltalk. But, there are no
>> implementors of #growMemoryByAtLeast: in the image..!
>
>
> If you download a recent Spur image from my web site you'll find it
> there-in.
>
>>
>>
>> Do you have an impl. of that method? This could be a real problem for
>> me to move to Spur because I absolutely depend on handling OutOfMemory
>> errors to build my large models..
>
>
>
>
> --
> best,
> Eliot

Eliot Miranda-2

Re: #'insufficient object memory' in Spur

On Fri, Nov 14, 2014 at 12:40 PM, Chris Muller <[hidden email]> wrote:

I was already using the 10/28 image. Ah ha! It seems I accidently
Load'ed that latest System package (unconverted) instead of Merge'ing
it.. OOPS! Not my day today!

Not your fault. I will autoupdate immediately.

On Fri, Nov 14, 2014 at 2:26 PM, Eliot Miranda <[hidden email]> wrote:
> Hi Chris,
>
> On Fri, Nov 14, 2014 at 7:43 AM, Chris Muller <[hidden email]> wrote:
>>
>> Hi Eliot, thanks a lot for your help on the CM header. I decided to
>> try to add a new bit indicator in my CompiledMethodBuffer (pickled
>> state of a CM) to indicate which header format it was pickled in so I
>> can hopefully maintain a heterogeneous environment.
>>
>> Now, maybe I goofed up my image while testing it at some point because
>> I cannot seem to allocate a 50MB ByteString (even though (Smalltalk
>> vmParameterAt: 3) reports just 89,627,904)
>>
>> String new: 50000000
>>
>> causes a failure in primitve 71, which causes the failure code to run,
>> which sends #growMemoryByAtLeast: to Smalltalk. But, there are no
>> implementors of #growMemoryByAtLeast: in the image..!
>
>
> If you download a recent Spur image from my web site you'll find it
> there-in.
>
>>
>>
>> Do you have an impl. of that method? This could be a real problem for
>> me to move to Spur because I absolutely depend on handling OutOfMemory
>> errors to build my large models..
>
>
>
>
> --
> best,
> Eliot

best,

Eliot