Marc,

I understand your point. I thought someone might make it.

However, lets look at this another way. If I tell about a databas manager I would like
to put in your office and it supports in memory tables. If you use in memory tables you
accept the risk that the data can be lost anytime up to actually shuting down the engine. I
guess most people would understand that, they would use them or not use them.
Now it also supports cached tables, meaning  the data is stored during runtime on disk.
At least that is what I took it to mean. One thing I tell you is that you must commit your
actions (call this a transaction) for the data to be saved to the table.
Now you start a transaction and you commit it, you receive no error. So where do you
expect your data to be? Well, I suggest most people expect it to be on disc.
But here it is not! IMO this is a real problem.

Marc Santhoff wrote:
Am Donnerstag, den 29.09.2005, 06:07 -0400 schrieb Andrew Jensen:
[...]
  
First this. Then there is the fact that when a connection is made to the 
datasource from
within basic module it is not automatically closed when the variable 
referencing it
goes out of scope. OK, I can see why that would be of course. But there 
will be
those that will create it and for one reason or a nother, like a 
programming error
never dispose of it.
    

I personally think a programmer should know what he or she is doing.

Assuming your idea of auto-closing would be implemented you'd have to
deal with the fact that you must re-open a connection any time you want
to use it. I think this would be even more confusing.

The main part here is a typically documentation thing, could you please
file an issue for making clear that a connection *has to be closed after
use* in the docs?

TIA,
Marc



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



  

Reply via email to