RDB introduction

[Wim Van den Wyngaert]

Didn't read all of it but this is a nice text explaining some stuff.
http://www.hpl.hp.com/techreports/Compaq-DEC/CRL-92-4.pdf

Wim

[Richard J Maher]

Ooh, and "indexes on functions of a column", how sexy is that! Is this on the way for Rdb or was end-of-development officially announced for Rdb somewher and I just missed it?

Also, how do you return a result-set from an Rdb stored procedure? With Packages Orrible has that MyDataSet type is Ref_Cursor syntax and both Oriible and Microsoft have the .NET class libraries that can handle it, but what is Rdb doing?

Cheers Richard Maher

[Jean-François Piéronne]

Richard,

you can't have "indexes on functions of a column" but adding a automatic column on your function and a index using this column has a similar result. I do this, also, for partitioning purpose.

I have found that the key index size limit more inconvenient.

JFP

[Wim Van den Wyngaert]

As far as I am now :
1) RDB has a superior locking strategy compared with Oracle and Sybase. If you only mark large read transaction as read only and all others as read/write snapshot overhead should be low and locking would be used elsewhere. Nice.

2) Too much possibilities to store dates/times. I still prefer the cobol way ccyymmddhhmmss.

3) Nice that DDL is not generating a commit but can be rolled back just as any statement.

4) Temporary tables are in VM. I prefer having it mapped to disk if the table is too big.

5) User access well integrated with VMS (no new user name at RDB level), access is ACL like and audit very VMS like.

6) I think all good features are planted in standard Oracle too (if it ever arrives on VMS).

Wim

[Richard J Maher]

Hi JFP,

"similar"? Like chalk is similar to chesse :-)

Have I got it wrong or does your Rdb strategy require two copies of the indexed data on disk (one for the column and one for the index)? Or perhaps I've misunderstood Orrible's and it too requires 2 on-disk column-data instances?

I perceive Orrible's offering as more akin to an index on an Rdb COMPUTED_BY column n'est pas?

All I was looking for was an upper-case version of a column that could be searched meaningfully and when I enquired weather Orrible's strategy was similar to a MySQL peculiarity of deploying sweedish_ci as a character set (or waving a dead-chicken over one's head) someone told me about indexing functions and I thought it was useful (from a diskspace point of view if nothing else)

As far as the more general question of "Rdb being actively developed - or not?", there are at least too extremely well paid "Technical Architects" on the payroll consuming copious license fees that I'm sure wouldn't be there if the product was in mothballs. So it's all good - I'm sure Rdb8 will be a cracker!

Cheers Richard Maher