Does DabbleDB use a database?

Does anyone know if DabbleDB uses a database at the back end?  Given the
speed of this application I'm inclined to believe that it's doesn't but I'm
not sure.  TIA.


John Treble
Ottawa, Ontario, Canada


_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

On May 17, 2006, at 4:29 PM, John Treble wrote:
>
> Does anyone know if DabbleDB uses a database at the back end?  
> Given the
> speed of this application I'm inclined to believe that it's doesn't  
> but I'm
> not sure.  TIA.

It doesn't use a relational database, no.  And it does very  
aggressively cache data in RAM while you're working with it (and  
equally aggressively swap it out when you're not).

Avi
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

What database does it use?
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

On 17-May-06, at 9:14 AM, Ramon Leon wrote:

Nothing third-party -- we've written all the transactional/
checkpointing/etc code for this ourselves, as nothing else did what  
we needed.. (not surprising as Dabble has a bit of an odd profile).

Andrew
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

I am not so sure that Dabble's profile is all that "odd" as you
put it. I think that persistence mechanism you guys have come up
with could be generally useful in a lot of scenarios. I for one
would very much like to understand the gory details of how you
went about attacking and solving the problem of persistence
in Dabble.  


----------=-=-=-=-=-=-=-=-========oOo========-=-=-=-=-=-=-=-=----------
mailto:[hidden email]                Phone:(607)770-3701
BAE SYSTEMS                     600 Main St Johnson City, NY 13790-1888
----------=-=-=-=-=-=-=-=-===================-=-=-=-=-=-=-=-=----------
-----Original Message-----
From: Andrew Catton [mailto:[hidden email]]
Sent: Wednesday, May 17, 2006 2:40 PM
To: The Squeak Enterprise Aubergines Server - general discussion.
Subject: Re: [Seaside] Does DabbleDB use a database?


On 17-May-06, at 9:14 AM, Ramon Leon wrote:

Nothing third-party -- we've written all the transactional/
checkpointing/etc code for this ourselves, as nothing else did what  
we needed.. (not surprising as Dabble has a bit of an odd profile).

Andrew
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

On 17-May-06, at 11:58 AM, Welch, Ronald P (US SSA) wrote:

> I am not so sure that Dabble's profile is all that "odd" as you
> put it. I think that persistence mechanism you guys have come up
> with could be generally useful in a lot of scenarios.

A few things make it odd IMO (you can of course judge how odd for  
yourself):
- the need to be able to reconfigure the schema on the fly, with  
existing data
- the fact that we're dealing with thousands of relatively small  
independent data stores
- the very incremental UI (causing us to, for example, display  
intermediate results as a query is built up rather than one result  
for a complex query built in one step)

Most DBs, relational and OO, are optimized more for a relatively  
static schema, and to be vertically scalable to allow one centralized  
data store to service many clients.

> I for one
> would very much like to understand the gory details of how you
> went about attacking and solving the problem of persistence
> in Dabble.

Fair enough.. we'll of course need to figure out which implementation  
details we want to publicly share, given we've joined this whole  
cloak-and-dagger commercial product thing :P  It probably isn't all  
that difficult to imagine the general approach we would take, but of  
course there were a lot of gotchas we needed to work through along  
the way in order to make it practical.

Cheers, Andrew
======================

Andrew Catton
Smallthought Systems Inc.
[hidden email]



_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

> A few things make it odd IMO (you can of course judge how odd for
> yourself):
> - the need to be able to reconfigure the schema on the fly,
> with existing data

I'd wondered about how you implemented this.

> Fair enough.. we'll of course need to figure out which
> implementation details we want to publicly share, given we've
> joined this whole cloak-and-dagger commercial product thing
> :P  It probably isn't all that difficult to imagine the
> general approach we would take, but of course there were a
> lot of gotchas we needed to work through along the way in
> order to make it practical.
>
> Cheers, Andrew

I'm also interested in anything you feel like sharing, storing what is
essentially an adaptive object model is an interesting problem.  I'd
assumed you were using an existing OO database, I didn't think you'd
write a custom one.
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside