Re: create package

Sybrand Bakker wrote:
>>we have a schema a, witch contains some packages that execute statistic
>>calculations and other things on some data and parameters stored in
>>tables in schema a.
>>
>>our customers need to "freeze" the state of the schema and restore it
>>later on. this is done by copying the rows in the tables to another
>>schema, b, and the packages are also saved to schema b, by copying the
>>source in a table.
>>
>>this can be done several times.
>>
>>we don't want to use sqlplus, because we don't want to exit on a shell.
>>
>>now we have this scenario: n packages and n parameter/data
>>configurations stored in schema b.
>>
>>the user can now choose one of these n configurations and put them back
>>to schema a.
>>
>>this works for the rows in the tables, but the packages need to be
>>restored also, and the limit of 32767 chars is killing me.

> 
> 
> 
> The obvious solution would be
> - develop *one* package
> - grant the users access to n schemas
> - arrange the access to the tables in your package  by setting up
> appropiate synonyms, drop and recreate them when access to a different
> set of packages is required
> 
> Evidently : no need at all to create packages on the fly
> 
> If you think that doesn't work for you there's something seriously
> wrong with your development methodology.

hi Bakker, tnx for your answer and the time you spend behind this thread. maybe i did not explain myself right.

we have this schema (s1), with some tables (t1) witch contain parameters, and some packages (p1) witch contain calculation algorythms. something like:
t1.parameter1 + t1.parameter2 / t1.parameter3

now, one day, our customers ask us to change this algorythm in something like
t1.parameter1 - t1.parameter2 * t1.parameter3.

befor we can do that, our customers can "archive" the current state of the schema. they do that, and anything in the production schema is "copyed" from s1, to s2 (our log schema). this thing has to be automated, of course. no manual intervention should me needed. so we have configuration1 in s2.

now we can change the algorythms in s1, p1.

the day after, our customers want to try the algorythms they used bevor, to look at the differences of the resoults.

doing what you sad (if i understood it right), just changes the references to the tables (t1) to some other tables (maybe those in configuration1 (s2)), but that does not change the used alghorythm in p1.

did i misunderstand something?

-- 

regards,
Simon