On Mon, 20 Aug 2001, Ian Holsman wrote:
> Sterling Hughes wrote:
>
> >
> >
> >>sorry bout that.
> >>MSVC does it's own thing with spacing ;(
> >>(ok.. I've run the code through indent so that should help.)
> >>
> >
> > It looks good -- a couple of comments:
> >
> > 1) Perhaps sync and optimize API functions could be added, yes,
> > some db's don't support this, but they can just do nothing for those
> > db's (no loss if there is no API for it).
> >
> sounds fair..
>
> >
> >
> > 2) Sometimes people may want to compile with support for two
> > databases and then decide at runtime which one they want to use.
> > Could this not be put into the current implementation?
> >
> yes.. thats the point of the patch
> #include <apr_dbm_sdbm.h>
> #include <apr_dbm_db.h>
>
> apr_dbm_t b1= apr_dbm_open_sdbm(...);
> apr_dbm_t b2= apr_dbm_open_db(...);
>
> should work.
> the 'test' program will have a update later which will allow it to
> choose DB types (-t DB|SDBM|GDBM)
>
Ahh, ok... I was thinking something more inline with:
#include <apr_dbm.h>
apr_dmb_t b1 = apr_dbm_open(APR_GDBM, ...);
But either way works.
> >
> > 3) Some of the Berkeley DB stuff doesn't look quite right to me,
> > however, I'm no expert (could be perfectly fine). Someone else should
> > probably take a closer look at this though....
> >
> I ran the test program on berkley 3.3.11 on NT, and the basic ops work
> (I don't think the current stuff does)
>
Ok, I took a look at the current implementation -- which seems even
iffier :)
-Sterling
