(domain support) Re: theory and practice: ying and yang

DM Unseen wrote:
> Paul,
>
> I cannot find anything on overloading SQL operators or functions with
> the newly made types in SQL server 2005
> and I'm not into FirstSQL.
>
> I do know that for SQL Server 2005 user defined types are actually
> special .NET objects that can live inside a database field (probably a
> BLOB field of some sort) by using .NET object serialization(i.e.
> converting the status info of an object to XML and probably compress
> the XML file and store it in a BLOB field). Defining such an object
> always requires *the developer* to write a special required
> method/property called something like ".totext" which returns a (human
> readable?) string representation of the object. This property needs
> then to be "exposed" to SQL server.
>
> Knowing this I fear SQL server will use this "totext" property for all
> operations like ordering/equality etc :(
>
> I'm not sure if I would call this extension of SQL server fully
> "native".(It is however fully Microsoft;)
>
> Hope this helps,
>
> Regards,
>
> M. Evers aka DM Unseen
>

hi M. Evers.

thanks for looking. i did a few searches but came up dry. it did sound a little too good to be true. perhaps i don't know the current M$ lingo well enough to search properly. the last time i used that stuff was when i took a course in mssql 6. i seem to recall that at that time, there was just a kind of blunt parser macro that let one of the "machine" types masquerade as a user type. not what i'd call full type support though. for example, my address can be written in several ways, and is, and the post office delivers them all to the same door. as far as i'm concerned those textually different addresses are equal, no matter what the machine bit configurations are. and different addresses could be ordered in several ways, depending on one's use of them.

given that two relations could be of the same type it all makes me wonder whether one could legimately define a different equality rule for RVA's, contrary (i think) to D&D. as for relative ordering, i remember reading somewhere that such comparisons "make no sense" but this objection has always seemed psychological to me, perhaps rooted in the bit comparisons dating back to the key-punch era.

paul c. Received on Thu May 26 2005 - 17:36:53 CEST