Marco Bambini wrote:It is safe to read the first N bytes in a db file and check for the string "** This file contains an SQLite 2..." or the string "** This file contains an SQLite 3..." to determine if the db was created with the 2.x or 3.x version?
Is there a better method?
I think that works. Though the correct prefixes are:
Version 2.8: [** This file contains an SQLite 2.1 database **] Version 3.0: [SQLite format 3.]
I propose that a simple validation function is added to both the 2.8 branch and the 3.0 branch of SQLite that someone can invoke like open() except that it simply returns true or false if the target file is a valid SQLite 2/3 file or not.
I would like a way for the SQLite library itself to tell me this in a simple and quick way; I don't want to have to first open a dubious file and then try to run SQL against it before I find out the file is invalid.
This function doesn't have to be thorough; it could simply check for both the above string and the magic number.
I think having to do such a check manually, opening a SQLite file without using the SQLite library, smacks of bad encapsulation.
Note that, unless the behaviour of open() itself is changed to do this, this function could be added after leaving beta stage as such an addition is backwards compatible. But it would be nice to have now.
One benefit of such a built-in is that I could use it from an application function that auto-detects SQLite database instances, so in case I just want to present the user a menu of local databases available for opening. In Perl, this can be used to help implement DBI's data_sources() function.
So shall I start a ticket for this?
-- Darren Duncan
P.S. Richard, thanks for posting 3.0.5. Matt, I yearn for your adaption of it.
