Re: [gnome-db] Access files

[Brian Bruns <camber ais org>]

On 27 Dec 2002, Rodrigo Moya wrote:

> On Fri, 2002-12-27 at 01:31, Brian Bruns wrote:
> > Do a google gov search for 'access 97 download' or something close.  It'll 
> > turn up a bunch of government databases in Access format, many with *gasp*
> > accompanying documentation! ;-)
> > 
> cool, thanks a lot for the info.
> 
> > By the way, I'm back from my hiatus and working on MDB Tools again and 
> > will be adding write support Real Soon Now (tm), along with some bug fixes 
> > and such.  Also the graphical tool (gmdb) has been ported to Gnome2 and 
> > libglade2.
> > 
> 
> this is great news!
> 
> BTW, I'm having problems with current CVS mdb-tools. I've got this code:
> 
> 	mdb_cnc = g_new0 (GdaMdbConnection, 1);
> 	mdb_cnc->cnc = cnc;
> 	mdb_cnc->server_version = NULL;
> 	mdb_cnc->mdb = mdb_open (filename);
> 	if (!mdb_cnc->mdb) {
> 		gda_connection_add_error_string (cnc, _("Could not open file %s"),
> filename);
> 		g_free (mdb_cnc);
> 		return FALSE;
> 	}
> 
> 	mdb_read_catalog (mdb_cnc->mdb, MDB_TABLE);
> 
> when running this code, when making the call to mdb_read_catalog, I get:
> 
> 0040  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  

>                                                     
> usage map found on page 6 start 64 end 2047
> Segmentation fault
> 
> any idea what's happening?

You may want to refresh from CVS, I'm in pretty heavy development right 
now since I'm off work until after new years.  I'm trying to keep my 
checkins functional, but not always successfully. ;-)

For the curious, the above garbage is a page allocation map which we are 
now using to scan the database instead of a brute force read, which makes 
large databases tremendously faster.  It looks like when you got your copy 
out of CVS, one style of usage map was supported and the second style
(indirect blocks) was not and thus the segfault. 

If re-uping from cvs doesn't work, let me know, I'm keen to get any bugs 
ironed out with it.
 
Cheers,

Brian

Oh, and by the way, once support for writing has been added I'll be 
looking at refactoring the API a bit, there will be a stable 0.5 release 
before that though.

> cheers
>