Smalltalk › Pharo › Pharo Smalltalk Developers

[for testing] speeding up code loading

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

6 messages Options

Marcus Denker-4

Oct 20, 2014; 3:48pm

[for testing] speeding up code loading

7582 posts

Hi,

[Object compileSilently: 'foo ^ 1' classified: 'testing'] bench

-- 267 per second

Load from inbox:

SLICE-Issue-14242-Installing-a-method-is-slow-MarcusDenker.3

run bench again ==> 2320 per second.

This will work as long the .changes is small enough so that it uses a normal

(not extended) source pointer.

Marcus

Esteban A. Maringolo

Oct 20, 2014; 5:44pm

Re: [for testing] speeding up code loading

2343 posts

2014-10-20 12:50 GMT-03:00 Marcus Denker <[hidden email]>:

> Hi,
>
> [Object compileSilently: 'foo ^ 1' classified: 'testing'] bench
>
> -- 267 per second
>
> Load from inbox:
> SLICE-Issue-14242-Installing-a-method-is-slow-MarcusDenker.3
>
> run bench again ==> 2320 per second.
>
> This will work as long the .changes is small enough so that it uses a
> normal
> (not extended) source pointer.

What is "small enough"?

How can I compress the changes file?

Regards!

Sven Van Caekenberghe-2

Oct 20, 2014; 6:06pm

Re: [for testing] speeding up code loading

5697 posts

> On 20 Oct 2014, at 19:45, Esteban A. Maringolo <[hidden email]> wrote:
>
> 2014-10-20 12:50 GMT-03:00 Marcus Denker <[hidden email]>:
>> Hi,
>>
>> [Object compileSilently: 'foo ^ 1' classified: 'testing'] bench
>>
>> -- 267 per second
>>
>> Load from inbox:
>> SLICE-Issue-14242-Installing-a-method-is-slow-MarcusDenker.3
>>
>> run bench again ==> 2320 per second.
>>
>> This will work as long the .changes is small enough so that it uses a
>> normal
>> (not extended) source pointer.
>
>
> What is "small enough"?
>
> How can I compress the changes file?

... [show rest of quote]

PharoChangesCondenser

> Regards!

Marcus Denker-5

Oct 20, 2014; 6:08pm

Re: [for testing] speeding up code loading

16 posts

In reply to this post by Esteban A. Maringolo

On Monday, Oct 20, 2014 at 7:46 pm, Esteban A. Maringolo <[hidden email]>, wrote:

2014-10-20 12:50 GMT-03:00 Marcus Denker :

> Hi,
>
> [Object compileSilently: 'foo ^ 1' classified: 'testing'] bench
>
> -- 267 per second
>
> Load from inbox:
> SLICE-Issue-14242-Installing-a-method-is-slow-MarcusDenker.3
>
> run bench again ==> 2320 per second.
>
> This will work as long the .changes is small enough so that it uses a
> normal
> (not extended) source pointer.

What is "small enough"?

@inria.fr>

... [show rest of quote]

some 32MB

How can I compress the changes file?
@inria.fr>

PharoChangesCondenser condense.

Ceterum censeo: we need to get rid of sources/changes…

Marcus

Sven Van Caekenberghe-2

Oct 20, 2014; 6:40pm

Re: [for testing] speeding up code loading

5697 posts

> On 20 Oct 2014, at 20:09, Marcus Denker <[hidden email]> wrote:
>
> Ceterum censeo: we need to get rid of sources/changes…

Now I learned another Latin proverb, nice !

Marcus Denker-4

Oct 21, 2014; 5:51am

Re: [for testing] speeding up code loading

7582 posts

In reply to this post by Marcus Denker-4

This is in Pharo4 update #314.

On Mon, Oct 20, 2014 at 5:50 PM, Marcus Denker <[hidden email]> wrote:

Hi,

[Object compileSilently: 'foo ^ 1' classified: 'testing'] bench

-- 267 per second

Load from inbox:

SLICE-Issue-14242-Installing-a-method-is-slow-MarcusDenker.3

run bench again ==> 2320 per second.

This will work as long the .changes is small enough so that it uses a normal

(not extended) source pointer.

Marcus

... [show rest of quote]