Re: [Firebird-devel] Re: Y-valve & aliases

[John B. <bel...@cs...>]

Hi all,

On Friday, June 8, 2001, at 01:52 PM, Paul Schmidt wrote:

> Sean:
>
> On 8 Jun 2001, at 13:52, Leyne, Sean wrote:
>
>> [......]
>> Finally, I'm curious -- Does anyone think that they would actual *use*
>> client side aliases?
>
> I think client side aliases might just carry a price, we don't want
> to have to pay for forever more, because we can't guarantee that a
> client side alias file is secure even if it were in a binary format
> and encrypted, it could still be deleted or replaced, and then you
> need programs to maintain it, it can't be cached and would increase
> the startup time.  Server side aliases are secured by the server and
> the network,  the server can cache the table (since it is always
> needing it).  My personal opinion is that if you allow users to
> create their own "private" aliases in the system alias table then we
> don't need client side aliases.

It is try that aliases carry a price.  But it should be possible to 
mitigate that price.  For example we could provide a configuration 
option that would disable client/server side alias expansion.  Heavy 
connection turnover databases wouldn't pay the price.  But my in-house 
MRP system that people connected to in the morning and disconnected from 
before they left in at night could use them.  Client side aliases would 
only be as secure as the user responsible for maintaining the file.  (As 
per the discussion before) if security is the intent, then only use 
server side aliases.  Otherwise client side aliases would be quite nice, 
just not on my external web server :)

>
> Now here is a point, the aliases should be maintained in a server
> wide database (similar to or even in isc4.gdb)
> [....]

-John