Maxim Sloyko wrote:
Well, I mostly do Web apps, so it is not a problem since user can't see code. If the user can see the code, can find a password int it and understand what password is this, then there is no way encryption can help much (storing keys etc. as you pointed it out.), because this kind of user can figure it all out, if he wants to.
The encrypted source code seems to solve that. I've seen it used for commercial apps. Basically you see the source filter plain text in a 'use' statement and the rest is jibberish. If you write the encrypting source filter in C and follow other guidlines it becomes next to impossible to know how to decrypt things. 'man perlfilter' if you want to take a look.
But this is not the point. The point was that usage of some file with passwords by *DEFAULT* is not the way to go, IMHO. It raises more problems than it solves.
I agree. We don't have any connect strings in code. We use higher level names that map to connect strings, subclassing DBI to do that. An advantage of that is that we can change what's under a logical connect name in one place (e.g., if a database moves). That puts them all in one place, which is a start anyway. What might make sense is some way to subclass and override just the connect string related access to make this sort of thing easier. -- Steve Sapovits [EMAIL PROTECTED]
