Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Eclipse Projects » DTP » SQL model containment from WTP/RDB
SQL model containment from WTP/RDB [message #855] Tue, 12 July 2005 14:31 Go to next message
Igor Jacy Lino Campista is currently offline Igor Jacy Lino CampistaFriend
Messages: 34
Registered: July 2009
Member
Hi,

I have seen the SQL model of the RDB package in the Rational Rose mdl file
and I have doubts about the containment relations between Database and
Schema, Schema and all others(i.e. Tables, Indexes...).

The current model as of now, when generated using the EMF, and using a
sqlschema resource (for a Schema resource or a Database resouce), it is
not possible to create a child 'Schema' in a Database, and in a Schema we
cannot create a child 'table'(and quite other important SQLObjects). As
this would be sound natural to do.

And because of this, its necessary to have separate resources and refer
them loading the resource. Which of course allows reusing of resources,
but should be merely an option that could be used.

In the SQL mdl model I have made a small change to the containment
relations from "contains a" to "creates a" and now everything works as
before, but now I can create a child 'Schema' in the Database, and as well
a child 'Table' in a Schema. [of course other types of children are
available and this is just the simplest examples].

Could there be an official modification to the mdl model concerning the
containment relations to allow such resource functionality?

Best regards,
Igor Lino
Re: SQL model containment from WTP/RDB [message #884 is a reply to message #855] Wed, 13 July 2005 14:10 Go to previous messageGo to next message
Eclipse UserFriend
Originally posted by: eclipse.ambysoft.com

At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
more sophisticated approach to UML data modeling than is currently
supported by Rose.

I suspect the problems you're talking about stem from the limitations
of that tool, so perhaps it's not the best thing to be looking at
right now as an example of what the DTP effort should be doing.

If you have any feedback regarding the profile, I'd love to hear it.

- Scott
On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
wrote:

>Hi,
>
>I have seen the SQL model of the RDB package in the Rational Rose mdl file
>and I have doubts about the containment relations between Database and
>Schema, Schema and all others(i.e. Tables, Indexes...).
>
>The current model as of now, when generated using the EMF, and using a
>sqlschema resource (for a Schema resource or a Database resouce), it is
>not possible to create a child 'Schema' in a Database, and in a Schema we
>cannot create a child 'table'(and quite other important SQLObjects). As
>this would be sound natural to do.
>
>And because of this, its necessary to have separate resources and refer
>them loading the resource. Which of course allows reusing of resources,
>but should be merely an option that could be used.
>
>In the SQL mdl model I have made a small change to the containment
>relations from "contains a" to "creates a" and now everything works as
>before, but now I can create a child 'Schema' in the Database, and as well
>a child 'Table' in a Schema. [of course other types of children are
>available and this is just the simplest examples].
>
>Could there be an official modification to the mdl model concerning the
>containment relations to allow such resource functionality?
>
>Best regards,
>Igor Lino
Re: SQL model containment from WTP/RDB [message #972 is a reply to message #884] Thu, 14 July 2005 07:39 Go to previous messageGo to next message
Igor Jacy Lino Campista is currently offline Igor Jacy Lino CampistaFriend
Messages: 34
Registered: July 2009
Member
Hi Scott,

Interesting profile, I will take a deeper look as I get eventually more
free time.

My small problem is that the RDB model has for the 'Schema' entity an
'Aggregation' relationship to persistent table, etc. And the 'Database'
has an 'Aggregation' relationship to the 'Schema' entity. They should be
of 'Composition' type, so that the eclipse resource enables what I
described in my first news post. The tool has no problems.

Of course Its on my interest that this could be made to the official
model, so I can rely on it.

Rational Rose from IBM Rational (adquired by IBM sometime ago), has in
fact quite much functionality used the proper way. At the moment we have
been able to model everything we have need.

Best regards,
Igor

Scott W. Ambler wrote:

> At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
> more sophisticated approach to UML data modeling than is currently
> supported by Rose.

> I suspect the problems you're talking about stem from the limitations
> of that tool, so perhaps it's not the best thing to be looking at
> right now as an example of what the DTP effort should be doing.

> If you have any feedback regarding the profile, I'd love to hear it.

> - Scott
> On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
> wrote:

>>Hi,
>>
>>I have seen the SQL model of the RDB package in the Rational Rose mdl file
>>and I have doubts about the containment relations between Database and
>>Schema, Schema and all others(i.e. Tables, Indexes...).
>>
>>The current model as of now, when generated using the EMF, and using a
>>sqlschema resource (for a Schema resource or a Database resouce), it is
>>not possible to create a child 'Schema' in a Database, and in a Schema we
>>cannot create a child 'table'(and quite other important SQLObjects). As
>>this would be sound natural to do.
>>
>>And because of this, its necessary to have separate resources and refer
>>them loading the resource. Which of course allows reusing of resources,
>>but should be merely an option that could be used.
>>
>>In the SQL mdl model I have made a small change to the containment
>>relations from "contains a" to "creates a" and now everything works as
>>before, but now I can create a child 'Schema' in the Database, and as well
>>a child 'Table' in a Schema. [of course other types of children are
>>available and this is just the simplest examples].
>>
>>Could there be an official modification to the mdl model concerning the
>>containment relations to allow such resource functionality?
>>
>>Best regards,
>>Igor Lino
Re: SQL model containment from WTP/RDB [message #1003 is a reply to message #972] Wed, 20 July 2005 20:26 Go to previous messageGo to next message
Brian Payton is currently offline Brian PaytonFriend
Messages: 154
Registered: July 2009
Senior Member
There was actually a lot of debate in the team that defined the SQL
Model regarding where to use composition and where not to. If I
remember right, one of the factors that was considered is the way EMF
persistence works. If you use composition rather than aggregation, when
you persist a parent object, all the composed children get persisted at
the same time in the same .xmi file. That was a problem for some people
building tools on top of the model.

I tend to prefer composition myself. In the part of the SQL Model I was
responsible for (the SQL Query Model, which models SQL
Select/Insert/Update/Delete statements), I used composition where
possible. (Persistence isn't a problem in the Query Model, since we use
SQL itself as the persistence mechanism.)

Brian Payton
SQL Tools Development
IBM

Igor Lino wrote:

> Hi Scott,
>
> Interesting profile, I will take a deeper look as I get eventually
> more free time.
>
> My small problem is that the RDB model has for the 'Schema' entity an
> 'Aggregation' relationship to persistent table, etc. And the
> 'Database' has an 'Aggregation' relationship to the 'Schema' entity.
> They should be of 'Composition' type, so that the eclipse resource
> enables what I described in my first news post. The tool has no problems.
>
> Of course Its on my interest that this could be made to the official
> model, so I can rely on it.
>
> Rational Rose from IBM Rational (adquired by IBM sometime ago), has in
> fact quite much functionality used the proper way. At the moment we
> have been able to model everything we have need.
>
> Best regards,
> Igor
>
> Scott W. Ambler wrote:
>
>> At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
>> more sophisticated approach to UML data modeling than is currently
>> supported by Rose.
>
>
>> I suspect the problems you're talking about stem from the limitations
>> of that tool, so perhaps it's not the best thing to be looking at
>> right now as an example of what the DTP effort should be doing.
>
>
>> If you have any feedback regarding the profile, I'd love to hear it.
>
>
>> - Scott
>> On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
>> wrote:
>
>
>>> Hi,
>>>
>>> I have seen the SQL model of the RDB package in the Rational Rose
>>> mdl file and I have doubts about the containment relations between
>>> Database and Schema, Schema and all others(i.e. Tables, Indexes...).
>>>
>>> The current model as of now, when generated using the EMF, and
>>> using a sqlschema resource (for a Schema resource or a Database
>>> resouce), it is not possible to create a child 'Schema' in a
>>> Database, and in a Schema we cannot create a child 'table'(and quite
>>> other important SQLObjects). As this would be sound natural to do.
>>>
>>> And because of this, its necessary to have separate resources and
>>> refer them loading the resource. Which of course allows reusing of
>>> resources, but should be merely an option that could be used.
>>>
>>> In the SQL mdl model I have made a small change to the containment
>>> relations from "contains a" to "creates a" and now everything works
>>> as before, but now I can create a child 'Schema' in the Database,
>>> and as well a child 'Table' in a Schema. [of course other types of
>>> children are available and this is just the simplest examples].
>>>
>>> Could there be an official modification to the mdl model concerning
>>> the containment relations to allow such resource functionality?
>>>
>>> Best regards,
>>> Igor Lino
>>
>
Re: SQL model containment from WTP/RDB [message #1060 is a reply to message #1003] Mon, 25 July 2005 08:30 Go to previous message
Igor Lino is currently offline Igor LinoFriend
Messages: 3
Registered: July 2009
Junior Member
Hi Brian,

Yes, the persistence seems to cause that. So now I'm wondering what could
be a good idea, if lets say we have a 'Database' that has 5 'Schema's with
200-300 'PersistentTable's and 50-100 'Column's per table. How does the
xmi resource would be efficiently handled. The current model would need
over 1000 .xmi files. Was there any alternative to have a containment on 1
xmi file describing the model of a single rdb database?

- Igor Lino
Re: SQL model containment from WTP/RDB [message #566022 is a reply to message #855] Wed, 13 July 2005 14:10 Go to previous message
eclipse is currently offline eclipseFriend
Messages: 19
Registered: July 2009
Junior Member
At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
more sophisticated approach to UML data modeling than is currently
supported by Rose.

I suspect the problems you're talking about stem from the limitations
of that tool, so perhaps it's not the best thing to be looking at
right now as an example of what the DTP effort should be doing.

If you have any feedback regarding the profile, I'd love to hear it.

- Scott
On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
wrote:

>Hi,
>
>I have seen the SQL model of the RDB package in the Rational Rose mdl file
>and I have doubts about the containment relations between Database and
>Schema, Schema and all others(i.e. Tables, Indexes...).
>
>The current model as of now, when generated using the EMF, and using a
>sqlschema resource (for a Schema resource or a Database resouce), it is
>not possible to create a child 'Schema' in a Database, and in a Schema we
>cannot create a child 'table'(and quite other important SQLObjects). As
>this would be sound natural to do.
>
>And because of this, its necessary to have separate resources and refer
>them loading the resource. Which of course allows reusing of resources,
>but should be merely an option that could be used.
>
>In the SQL mdl model I have made a small change to the containment
>relations from "contains a" to "creates a" and now everything works as
>before, but now I can create a child 'Schema' in the Database, and as well
>a child 'Table' in a Schema. [of course other types of children are
>available and this is just the simplest examples].
>
>Could there be an official modification to the mdl model concerning the
>containment relations to allow such resource functionality?
>
>Best regards,
>Igor Lino
Re: SQL model containment from WTP/RDB [message #566129 is a reply to message #884] Thu, 14 July 2005 07:39 Go to previous message
Igor Jacy Lino Campista is currently offline Igor Jacy Lino CampistaFriend
Messages: 34
Registered: July 2009
Member
Hi Scott,

Interesting profile, I will take a deeper look as I get eventually more
free time.

My small problem is that the RDB model has for the 'Schema' entity an
'Aggregation' relationship to persistent table, etc. And the 'Database'
has an 'Aggregation' relationship to the 'Schema' entity. They should be
of 'Composition' type, so that the eclipse resource enables what I
described in my first news post. The tool has no problems.

Of course Its on my interest that this could be made to the official
model, so I can rely on it.

Rational Rose from IBM Rational (adquired by IBM sometime ago), has in
fact quite much functionality used the proper way. At the moment we have
been able to model everything we have need.

Best regards,
Igor

Scott W. Ambler wrote:

> At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
> more sophisticated approach to UML data modeling than is currently
> supported by Rose.

> I suspect the problems you're talking about stem from the limitations
> of that tool, so perhaps it's not the best thing to be looking at
> right now as an example of what the DTP effort should be doing.

> If you have any feedback regarding the profile, I'd love to hear it.

> - Scott
> On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
> wrote:

>>Hi,
>>
>>I have seen the SQL model of the RDB package in the Rational Rose mdl file
>>and I have doubts about the containment relations between Database and
>>Schema, Schema and all others(i.e. Tables, Indexes...).
>>
>>The current model as of now, when generated using the EMF, and using a
>>sqlschema resource (for a Schema resource or a Database resouce), it is
>>not possible to create a child 'Schema' in a Database, and in a Schema we
>>cannot create a child 'table'(and quite other important SQLObjects). As
>>this would be sound natural to do.
>>
>>And because of this, its necessary to have separate resources and refer
>>them loading the resource. Which of course allows reusing of resources,
>>but should be merely an option that could be used.
>>
>>In the SQL mdl model I have made a small change to the containment
>>relations from "contains a" to "creates a" and now everything works as
>>before, but now I can create a child 'Schema' in the Database, and as well
>>a child 'Table' in a Schema. [of course other types of children are
>>available and this is just the simplest examples].
>>
>>Could there be an official modification to the mdl model concerning the
>>containment relations to allow such resource functionality?
>>
>>Best regards,
>>Igor Lino
Re: SQL model containment from WTP/RDB [message #566144 is a reply to message #972] Wed, 20 July 2005 20:26 Go to previous message
Brian Payton is currently offline Brian PaytonFriend
Messages: 154
Registered: July 2009
Senior Member
There was actually a lot of debate in the team that defined the SQL
Model regarding where to use composition and where not to. If I
remember right, one of the factors that was considered is the way EMF
persistence works. If you use composition rather than aggregation, when
you persist a parent object, all the composed children get persisted at
the same time in the same .xmi file. That was a problem for some people
building tools on top of the model.

I tend to prefer composition myself. In the part of the SQL Model I was
responsible for (the SQL Query Model, which models SQL
Select/Insert/Update/Delete statements), I used composition where
possible. (Persistence isn't a problem in the Query Model, since we use
SQL itself as the persistence mechanism.)

Brian Payton
SQL Tools Development
IBM

Igor Lino wrote:

> Hi Scott,
>
> Interesting profile, I will take a deeper look as I get eventually
> more free time.
>
> My small problem is that the RDB model has for the 'Schema' entity an
> 'Aggregation' relationship to persistent table, etc. And the
> 'Database' has an 'Aggregation' relationship to the 'Schema' entity.
> They should be of 'Composition' type, so that the eclipse resource
> enables what I described in my first news post. The tool has no problems.
>
> Of course Its on my interest that this could be made to the official
> model, so I can rely on it.
>
> Rational Rose from IBM Rational (adquired by IBM sometime ago), has in
> fact quite much functionality used the proper way. At the moment we
> have been able to model everything we have need.
>
> Best regards,
> Igor
>
> Scott W. Ambler wrote:
>
>> At www.agiledata.org/essays/umlDataModelingProfile.html I have a much
>> more sophisticated approach to UML data modeling than is currently
>> supported by Rose.
>
>
>> I suspect the problems you're talking about stem from the limitations
>> of that tool, so perhaps it's not the best thing to be looking at
>> right now as an example of what the DTP effort should be doing.
>
>
>> If you have any feedback regarding the profile, I'd love to hear it.
>
>
>> - Scott
>> On Tue, 12 Jul 2005 14:31:11 +0000 (UTC), icampista@gmx.de (Igor Lino)
>> wrote:
>
>
>>> Hi,
>>>
>>> I have seen the SQL model of the RDB package in the Rational Rose
>>> mdl file and I have doubts about the containment relations between
>>> Database and Schema, Schema and all others(i.e. Tables, Indexes...).
>>>
>>> The current model as of now, when generated using the EMF, and
>>> using a sqlschema resource (for a Schema resource or a Database
>>> resouce), it is not possible to create a child 'Schema' in a
>>> Database, and in a Schema we cannot create a child 'table'(and quite
>>> other important SQLObjects). As this would be sound natural to do.
>>>
>>> And because of this, its necessary to have separate resources and
>>> refer them loading the resource. Which of course allows reusing of
>>> resources, but should be merely an option that could be used.
>>>
>>> In the SQL mdl model I have made a small change to the containment
>>> relations from "contains a" to "creates a" and now everything works
>>> as before, but now I can create a child 'Schema' in the Database,
>>> and as well a child 'Table' in a Schema. [of course other types of
>>> children are available and this is just the simplest examples].
>>>
>>> Could there be an official modification to the mdl model concerning
>>> the containment relations to allow such resource functionality?
>>>
>>> Best regards,
>>> Igor Lino
>>
>
Re: SQL model containment from WTP/RDB [message #566188 is a reply to message #1003] Mon, 25 July 2005 08:30 Go to previous message
Igor Lino is currently offline Igor LinoFriend
Messages: 3
Registered: July 2009
Junior Member
Hi Brian,

Yes, the persistence seems to cause that. So now I'm wondering what could
be a good idea, if lets say we have a 'Database' that has 5 'Schema's with
200-300 'PersistentTable's and 50-100 'Column's per table. How does the
xmi resource would be efficiently handled. The current model would need
over 1000 .xmi files. Was there any alternative to have a containment on 1
xmi file describing the model of a single rdb database?

- Igor Lino
Previous Topic:Draft of DTP Project Plan Posted
Next Topic:[rdb] DDL generate for one column only
Goto Forum:
  


Current Time: Thu Mar 28 08:20:09 GMT 2024

Powered by FUDForum. Page generated in 0.02525 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top