springframework-developer

Re: [Springframework-developer] src for long keys

[Thomas R. <tri...@tr...>]

Jean-Pierre,

Thanks for your changes.  I think adding long as a return value has been
 planned, but it was never implemented.  Isabelle, do you you want to
make these changes or do you want me to do it?  I'm modifying a few
other jdbc classes this weekend, so I don't mind adding these changes as
well.

For key generation support I think we should provide support for the
following:

1.  Sequences - like the OracleSequenceMaxValueIncrementer.  Do we need
something for Postgres and DB2?

2.  Identity/Autoincrement - a "sequence" table where we use some sql to
increment and retrieve the next value.  This is implemented in
MySQLMaxValueIncrementer.  Do we need implementations for HSQL, MS SQL
Server, DB2 and Sybase?

3.  Stored Procedure = a procedure that wraps access to a Sequence,
Identity/Autoincrement or any other proprietary database key generation
facility.  I'll look into providing an implementation for this.

I think we should discourage use of "select max(id) + 1" type of
solutions and therefore we should remove the RdbmsMaxValueIncrementer. 
We can leave this type of implementation up to the individual developer.

Thomas

> This is a multi-part message in MIME format.
> 
> 
> Hi Isabelle,
> 
> Related on my last post on the Spring Framework list, I enclosed my
> proposals of changes to handle long keys on databases.
> As I'm not registered as developer, I cannot upload. And in all cases,
> it's your work, so your approval is also mandatory.
> 
> NB: It's nothing about caching in this proposal.
> 
> Best regards,
> Jean-Pierre PAWLAK
> jp....@ti...
>  

Re: [Springframework-developer] src for long keys

[Thomas R. <tri...@tr...>]

Isabelle,

For NUMERIC we would be better off mapping to BigDecimal and 
use .longValue() or just use getLong().  Floats and Doubles are not 
reliably re-cast to a long for very large numbers so their use is 
limited.

I don't see why we would need to support anything other that int, long 
and String as return values.

Thomas

P.S.

I think you just replied to my e-mail address and not the list's - 
here is the message:

> Hi Thomas, Jean-Pierre,
> 
> If you don't mind adding Jean-Pierre's code, please do. I'm working 
on some code samples for the tutorial.
> I agree that Long's make sense, I was going to do it but then it 
slipped my mind.
> 
> The Double return value is there because of the practice of using 
NUMERIC columns for sequences.
> 
> As for retrieving bunches of keys at once, and then adding them out 
one by one, I agree that that makes sense as well. As a matter of fact 
it was raised in the mailing list when we first talked about insert 
functionality and key generation, I remember I referred to an open 
source project that does just that. It's not that complicated to do. 
Jean-Pierre, if you feel like doing it, why don't you go ahead, and 
then either Rod can give you commit access or one of us can check it 
in for you.
> 
> I completely agree on select max(id) + 1, it's very bad. I put in a 
code comment that explains why, but it's better than nothing for 
databases we do not have specific code for.
> 
> Isabelle
> 
> On Sat, May 03, 2003 at 11:36:18AM -0400, Thomas Risberg wrote:
> > Jean-Pierre,
> > 
> > Thanks for your changes.  I think adding long as a return value 
has been
> >  planned, but it was never implemented.  Isabelle, do you you want 
to
> > make these changes or do you want me to do it?  I'm modifying a few
> > other jdbc classes this weekend, so I don't mind adding these 
changes as
> > well.
> > 
> > For key generation support I think we should provide support for 
the
> > following:
> > 
> > 1.  Sequences - like the OracleSequenceMaxValueIncrementer.  Do we 
need
> > something for Postgres and DB2?
> > 
> > 2.  Identity/Autoincrement - a "sequence" table where we use some 
sql to
> > increment and retrieve the next value.  This is implemented in
> > MySQLMaxValueIncrementer.  Do we need implementations for HSQL, MS 
SQL
> > Server, DB2 and Sybase?
> > 
> > 3.  Stored Procedure = a procedure that wraps access to a Sequence,
> > Identity/Autoincrement or any other proprietary database key 
generation
> > facility.  I'll look into providing an implementation for this.
> > 
> > I think we should discourage use of "select max(id) + 1" type of
> > solutions and therefore we should remove the 
RdbmsMaxValueIncrementer. 
> > We can leave this type of implementation up to the individual 
developer.
> > 
> > Thomas
> > 
> > > This is a multi-part message in MIME format.
> > > 
> > > 
> > > Hi Isabelle,
> > > 
> > > Related on my last post on the Spring Framework list, I enclosed 
my
> > > proposals of changes to handle long keys on databases.
> > > As I'm not registered as developer, I cannot upload. And in all 
cases,
> > > it's your work, so your approval is also mandatory.
> > > 
> > > NB: It's nothing about caching in this proposal.
> > > 
> > > Best regards,
> > > Jean-Pierre PAWLAK
> > > jp....@ti...
> > >  
> > 
> > 
> > 
> > -------------------------------------------------------
> > This sf.net email is sponsored by:ThinkGeek
> > Welcome to geek heaven.
> > http://thinkgeek.com/sf
> > _______________________________________________
> > Springframework-developer mailing list
> > Spr...@li...
> > https://lists.sourceforge.net/lists/listinfo/springframework-
developer
> > 
> > 
> 
> -- 
> Isabelle Muszynski
> Software Engineer
> Zandweellaan 4
> 2660 Antwerpen
> Belgium
> Tel. 32-(0)3-830 18 54
> Mobile: 32-(0)485 49 50 89
> Email: isa...@me...
> Website: www.meta-logix.com
> 
> 


> Jean-Pierre,
> 
> Thanks for your changes.  I think adding long as a return value has 
been
>  planned, but it was never implemented.  Isabelle, do you you want to
> make these changes or do you want me to do it?  I'm modifying a few
> other jdbc classes this weekend, so I don't mind adding these 
changes as
> well.
> 
> For key generation support I think we should provide support for the
> following:
> 
> 1.  Sequences - like the OracleSequenceMaxValueIncrementer.  Do we 
need
> something for Postgres and DB2?
> 
> 2.  Identity/Autoincrement - a "sequence" table where we use some 
sql to
> increment and retrieve the next value.  This is implemented in
> MySQLMaxValueIncrementer.  Do we need implementations for HSQL, MS 
SQL
> Server, DB2 and Sybase?
> 
> 3.  Stored Procedure = a procedure that wraps access to a Sequence,
> Identity/Autoincrement or any other proprietary database key 
generation
> facility.  I'll look into providing an implementation for this.
> 
> I think we should discourage use of "select max(id) + 1" type of
> solutions and therefore we should remove the 
RdbmsMaxValueIncrementer. 
> We can leave this type of implementation up to the individual 
developer.
> 
> Thomas
> 
> > This is a multi-part message in MIME format.
> > 
> > 
> > Hi Isabelle,
> > 
> > Related on my last post on the Spring Framework list, I enclosed my
> > proposals of changes to handle long keys on databases.
> > As I'm not registered as developer, I cannot upload. And in all 
cases,
> > it's your work, so your approval is also mandatory.
> > 
> > NB: It's nothing about caching in this proposal.
> > 
> > Best regards,
> > Jean-Pierre PAWLAK
> > jp....@ti...
> >  
> 
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Springframework-developer mailing list
> Spr...@li...
> https://lists.sourceforge.net/lists/listinfo/springframework-
developer
> 
> 

-- 
Thomas Risberg
tri...@tr...

Re: [Springframework-developer] src for long keys

[Isabelle M. <isa...@me...>]

Hi Thomas,

I guess you're right. Even if someone declares the key as a NUMBER, it makes no sense to retrieve it as a Double. 

Isabelle

On Sat, May 03, 2003 at 02:29:20PM -0400, Thomas Risberg wrote:
> Isabelle,
> 
> For NUMERIC we would be better off mapping to BigDecimal and 
> use .longValue() or just use getLong().  Floats and Doubles are not 
> reliably re-cast to a long for very large numbers so their use is 
> limited.
> 
> I don't see why we would need to support anything other that int, long 
> and String as return values.
> 
> Thomas
> 
> P.S.
> 
> I think you just replied to my e-mail address and not the list's - 
> here is the message:
> 
> > Hi Thomas, Jean-Pierre,
> > 
> > If you don't mind adding Jean-Pierre's code, please do. I'm working 
> on some code samples for the tutorial.
> > I agree that Long's make sense, I was going to do it but then it 
> slipped my mind.
> > 
> > The Double return value is there because of the practice of using 
> NUMERIC columns for sequences.
> > 
> > As for retrieving bunches of keys at once, and then adding them out 
> one by one, I agree that that makes sense as well. As a matter of fact 
> it was raised in the mailing list when we first talked about insert 
> functionality and key generation, I remember I referred to an open 
> source project that does just that. It's not that complicated to do. 
> Jean-Pierre, if you feel like doing it, why don't you go ahead, and 
> then either Rod can give you commit access or one of us can check it 
> in for you.
> > 
> > I completely agree on select max(id) + 1, it's very bad. I put in a 
> code comment that explains why, but it's better than nothing for 
> databases we do not have specific code for.
> > 
> > Isabelle
> > 
> > On Sat, May 03, 2003 at 11:36:18AM -0400, Thomas Risberg wrote:
> > > Jean-Pierre,
> > > 
> > > Thanks for your changes.  I think adding long as a return value 
> has been
> > >  planned, but it was never implemented.  Isabelle, do you you want 
> to
> > > make these changes or do you want me to do it?  I'm modifying a few
> > > other jdbc classes this weekend, so I don't mind adding these 
> changes as
> > > well.
> > > 
> > > For key generation support I think we should provide support for 
> the
> > > following:
> > > 
> > > 1.  Sequences - like the OracleSequenceMaxValueIncrementer.  Do we 
> need
> > > something for Postgres and DB2?
> > > 
> > > 2.  Identity/Autoincrement - a "sequence" table where we use some 
> sql to
> > > increment and retrieve the next value.  This is implemented in
> > > MySQLMaxValueIncrementer.  Do we need implementations for HSQL, MS 
> SQL
> > > Server, DB2 and Sybase?
> > > 
> > > 3.  Stored Procedure = a procedure that wraps access to a Sequence,
> > > Identity/Autoincrement or any other proprietary database key 
> generation
> > > facility.  I'll look into providing an implementation for this.
> > > 
> > > I think we should discourage use of "select max(id) + 1" type of
> > > solutions and therefore we should remove the 
> RdbmsMaxValueIncrementer. 
> > > We can leave this type of implementation up to the individual 
> developer.
> > > 
> > > Thomas
> > > 
> > > > This is a multi-part message in MIME format.
> > > > 
> > > > 
> > > > Hi Isabelle,
> > > > 
> > > > Related on my last post on the Spring Framework list, I enclosed 
> my
> > > > proposals of changes to handle long keys on databases.
> > > > As I'm not registered as developer, I cannot upload. And in all 
> cases,
> > > > it's your work, so your approval is also mandatory.
> > > > 
> > > > NB: It's nothing about caching in this proposal.
> > > > 
> > > > Best regards,
> > > > Jean-Pierre PAWLAK
> > > > jp....@ti...
> > > >  
> > > 
> > > 
> > > 
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:ThinkGeek
> > > Welcome to geek heaven.
> > > http://thinkgeek.com/sf
> > > _______________________________________________
> > > Springframework-developer mailing list
> > > Spr...@li...
> > > https://lists.sourceforge.net/lists/listinfo/springframework-
> developer
> > > 
> > > 
> > 
> > -- 
> > Isabelle Muszynski
> > Software Engineer
> > Zandweellaan 4
> > 2660 Antwerpen
> > Belgium
> > Tel. 32-(0)3-830 18 54
> > Mobile: 32-(0)485 49 50 89
> > Email: isa...@me...
> > Website: www.meta-logix.com
> > 
> > 
> 
> 
> > Jean-Pierre,
> > 
> > Thanks for your changes.  I think adding long as a return value has 
> been
> >  planned, but it was never implemented.  Isabelle, do you you want to
> > make these changes or do you want me to do it?  I'm modifying a few
> > other jdbc classes this weekend, so I don't mind adding these 
> changes as
> > well.
> > 
> > For key generation support I think we should provide support for the
> > following:
> > 
> > 1.  Sequences - like the OracleSequenceMaxValueIncrementer.  Do we 
> need
> > something for Postgres and DB2?
> > 
> > 2.  Identity/Autoincrement - a "sequence" table where we use some 
> sql to
> > increment and retrieve the next value.  This is implemented in
> > MySQLMaxValueIncrementer.  Do we need implementations for HSQL, MS 
> SQL
> > Server, DB2 and Sybase?
> > 
> > 3.  Stored Procedure = a procedure that wraps access to a Sequence,
> > Identity/Autoincrement or any other proprietary database key 
> generation
> > facility.  I'll look into providing an implementation for this.
> > 
> > I think we should discourage use of "select max(id) + 1" type of
> > solutions and therefore we should remove the 
> RdbmsMaxValueIncrementer. 
> > We can leave this type of implementation up to the individual 
> developer.
> > 
> > Thomas
> > 
> > > This is a multi-part message in MIME format.
> > > 
> > > 
> > > Hi Isabelle,
> > > 
> > > Related on my last post on the Spring Framework list, I enclosed my
> > > proposals of changes to handle long keys on databases.
> > > As I'm not registered as developer, I cannot upload. And in all 
> cases,
> > > it's your work, so your approval is also mandatory.
> > > 
> > > NB: It's nothing about caching in this proposal.
> > > 
> > > Best regards,
> > > Jean-Pierre PAWLAK
> > > jp....@ti...
> > >  
> > 
> > 
> > 
> > -------------------------------------------------------
> > This sf.net email is sponsored by:ThinkGeek
> > Welcome to geek heaven.
> > http://thinkgeek.com/sf
> > _______________________________________________
> > Springframework-developer mailing list
> > Spr...@li...
> > https://lists.sourceforge.net/lists/listinfo/springframework-
> developer
> > 
> > 
> 
> -- 
> Thomas Risberg
> tri...@tr...
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Springframework-developer mailing list
> Spr...@li...
> https://lists.sourceforge.net/lists/listinfo/springframework-developer
> 
> 

-- 
Isabelle Muszynski
Software Engineer
Zandweellaan 4
2660 Antwerpen
Belgium
Tel. 32-(0)3-830 18 54
Mobile: 32-(0)485 49 50 89
Email: isa...@me...
Website: www.meta-logix.com

Re: RE : [Springframework-developer] src for long keys

[Isabelle M. <isa...@me...>]

Hello Jean-Pierre,

It might be a good idea to write an Oracle implementation as well, and then post to the list asking someone to test it. You are right, it is only a performance enhancement.

Isabelle


On Sat, May 03, 2003 at 09:29:36PM +0200, JP PAWLAK(Tiscali) wrote:
> Hi Isabelle,
> 
> So, my interrogation about the Double return key has now an answer. 
> I can implement the retrieving by bunches of keys. I better will not do
> for Oracle like RDBMS for which the increment factor is to retrieve from
> the sequence definition. Yes, it's only a matter of a SQL Select, but
> the time I was Oracle DBA is far and I will not have easy the
> possibility of testing it.
> That's said, it's not an urgent task as only a performance deal and I
> have work on implementing my application on the new framework with a lot
> of refactoring. Nevertheless, as key bunches are not a big deal, I hope
> work on in a few days.  
> 
> Regards,
> Jean-Pierre PAWLAK
> jp....@ti...
> 
> 
> -----Message d'origine-----
> De : spr...@li...
> [mailto:spr...@li...] De la
> part de Isabelle Muszynski
> Envoyé : samedi 3 mai 2003 20:44
> À : Thomas Risberg
> Cc : spr...@li...
> Objet : Re: [Springframework-developer] src for long keys
> 
> Hi Thomas,
> 
> I guess you're right. Even if someone declares the key as a NUMBER, it
> makes no sense to retrieve it as a Double. 
> 
> Isabelle
> 
> On Sat, May 03, 2003 at 02:29:20PM -0400, Thomas Risberg wrote:
> > Isabelle,
> > 
> > For NUMERIC we would be better off mapping to BigDecimal and 
> > use .longValue() or just use getLong().  Floats and Doubles are not 
> > reliably re-cast to a long for very large numbers so their use is 
> > limited.
> > 
> > I don't see why we would need to support anything other that int, long
> 
> > and String as return values.
> > 
> > Thomas
> > 
> > P.S.
> > 
> > I think you just replied to my e-mail address and not the list's - 
> > here is the message:
> > 
> > > Hi Thomas, Jean-Pierre,
> > > 
> > > If you don't mind adding Jean-Pierre's code, please do. I'm working 
> > on some code samples for the tutorial.
> > > I agree that Long's make sense, I was going to do it but then it 
> > slipped my mind.
> > > 
> > > The Double return value is there because of the practice of using 
> > NUMERIC columns for sequences.
> > > 
> > > As for retrieving bunches of keys at once, and then adding them out 
> > one by one, I agree that that makes sense as well. As a matter of fact
> 
> > it was raised in the mailing list when we first talked about insert 
> > functionality and key generation, I remember I referred to an open 
> > source project that does just that. It's not that complicated to do. 
> > Jean-Pierre, if you feel like doing it, why don't you go ahead, and 
> > then either Rod can give you commit access or one of us can check it 
> > in for you.
> > > 
> > > I completely agree on select max(id) + 1, it's very bad. I put in a 
> > code comment that explains why, but it's better than nothing for 
> > databases we do not have specific code for.
> > >
> 
>  
> ...
> 
> 
> 
> 

-- 
Isabelle Muszynski
Software Engineer
Zandweellaan 4
2660 Antwerpen
Belgium
Tel. 32-(0)3-830 18 54
Mobile: 32-(0)485 49 50 89
Email: isa...@me...
Website: www.meta-logix.com