Ever growing Pharo 2.0 image

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

Ever growing Pharo 2.0 image

hilaire
CONTENTS DELETED
The author has deleted this message.
Reply | Threaded
Open this post in threaded view
|

Re: Ever growing Pharo 2.0 image

Marcus Denker-4

On 29 Jan 2014, at 08:51, Hilaire Fernandes <[hidden email]> wrote:

> Hello,
>
> I have an ever growing image, now about 57MB and can't get it down:

Monticello is one problem:

        MCVersionInfo allInstances do: [ :each | each instVarNamed: 'ancestors' put: nil ].

this shrinks current Pharo3 by 5MB.

(but this does not explain all your memory use)

        Marcus
Reply | Threaded
Open this post in threaded view
|

Re: Ever growing Pharo 2.0 image

hilaire
CONTENTS DELETED
The author has deleted this message.
Reply | Threaded
Open this post in threaded view
|

Re: Ever growing Pharo 2.0 image

Marcus Denker-4
In reply to this post by hilaire

On 29 Jan 2014, at 08:51, Hilaire Fernandes <[hidden email]> wrote:

> Hello,
>
> I have an ever growing image, now about 57MB and can't get it down:
> - I reset the variables in workspace
> - purge my class instances
> - Check with SpaceTally my classes
> - do a ImageCleaner cleanUpForRelease
> - chack for ramping processes
>
> I mainly use Morph and Form, but I don't see them consuming space.
>
> Oh by the way:
> MorphExtension allInstances size. 7480
> Morph allSubInstances  size. 11184
>


-> closing all windows reduces the image to 37MB.

After that, doing a  spaceTally show this entry which is very strange: 177842 instances of TextMethodLink
taking 10% of memory:

TextMethodLink                                        419      177842       4268208        9.90               24.00

This is especially strange as this is only used in Shout syntax highlighting…

Maybe this is the bug we fixed in Pharo3 that made completion hold on to editors?

All in all, we did put quite some effort into finding memory leaks in Pharo3, some of which might have been there
allready in Pharo2.

        Marcus