difference between MaDictionary and MagmaDictionary

mmm...
AFAIU the objective of using magma collections is to avoid the read/write of big collections.... and I have this structure:

MaDictionary
       key -> MagmaCollection (big)
       key -> MagmaCollection (big)
       key -> MagmaCollection (big)

If I replace MaDictionary with Dictionary, I will be reading/writing the full dictionary each time I add an element to the dictionary... isn't?

btw... I'm using an unique MagmaSession to connect to repository (no real need of other... no more than 3 users connected at the same time)... and the real reason I'm asking this is because my application continues crashing each one or two days, and I can't figure out what's happening :(

So, any "best practice" I can use would be really welcome (I'm a little bit lost here... I'm new to magma in production, used just for toy project before this :( )

Cheers,
Esteban

El 14/12/2010, a las 6:12a.m., Igor Stasenko escribió:

> On 14 December 2010 01:40, Esteban Lorenzano <[hidden email]> wrote:
>> Hi,
>> I wonder what's the difference between MaDictionary and MagmaDictionary... and which should I use... is indistinct?
>>
>
> I think you better do not use these dicts. They are mainly for
> internal use only.
> They are optimized for speed not for space. And taking part in keeping
> correspondence between
> objects stored in database and their ids.
>
>> Cheers,
>> Esteban_______________________________________________
>> Magma mailing list
>> [hidden email]
>> http://lists.squeakfoundation.org/mailman/listinfo/magma
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.

_______________________________________________
Magma mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/magma

MagmaDictionary is just a normal MagmaCollection with an index on
#hash (actually, #hash32 because its just a 32-bit index).  Please see
it in a hierarchy browser.  As a subclass of MagmaCollection, it
inherits its characteristics...  Which is actually unfortunate because
those characteristics of regular MagmaCollections have highest
throughput when answering _batches_ of rows.  But dictionary's are
accessed by #at:, just one object at a time.

This is why I wrote MagmaPreallocatedDictionary.  It is based on a MagmaArray.

A MagmaArray is an auto-growing, but never shrinking, Array that can
be very large.  Please see the test cases whichh reference MagmaArray
for the actual semantics.

MagmaArrays are as fast as possible, always just two disk seeks,
6-bytes each, to get the location of the contents of any object.
That's why MagmaPreallocatedDictionary is much faster than a
MagmaCollection, which is based on a MaHashIndex file-structure for
supporting range-based querying.

 - Chris

On Tue, Dec 14, 2010 at 4:57 PM, Esteban Lorenzano <[hidden email]> wrote:
> ok, and what about using MagmaDictionary? better? or the same?
>
> cheers,
> Esteban
>
> El 14/12/2010, a las 5:40p.m., Chris Muller escribió:
>
>>>> If I replace MaDictionary with Dictionary, I will be reading/writing the
>>>> full dictionary each time I add an element to the dictionary... isn't?
>>
>> Yes, that's true.  But a MaDictionary only saves you that if you have
>> a lot of collisions.
>>
>>> I think that you need replace MaDictionary with MagmaPreallocatedDictionary
>>> not with standard Dictionary.
>>
>> No, this is wrong, please see how MagmaPreallocatedDictionary works.
>> It is for very large dictionary's, Esteban is using a MaDictionary for
>> his root object, and my guess is it only has a few entries in it..
>>
>> - Chris
>
> _______________________________________________
> Magma mailing list
> [hidden email]
> http://lists.squeakfoundation.org/mailman/listinfo/magma
>
_______________________________________________
Magma mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/magma