On Tue, Mar 29, 2005 at 11:46:26PM -0500, Stas Bekman wrote:
> take_imp_data support for DBD::mysql. it seems to be too simple,
The sign of a good solution... if it works :)
> Index: lib/DBD/mysql.pm
> - # Call msqlConnect func in mSQL.xs file
> - # and populate internal handle data.
> - DBD::mysql::db::_login($this, $dsn, $username, $password)
> - or $this = undef;
> + # if dbi_imp_data is passed then we should already have a handle,
> + # with a valid connection
> + unless ($privateAttrHash->{dbi_imp_data}) {
> + # Call msqlConnect func in mSQL.xs file
> + # and populate internal handle data.
> + DBD::mysql::db::_login($this, $dsn, $username, $password)
> + or $this = undef;
> + }
I think it's important to not change this code but put the "am I
reusing imp_data" logic into the drivers _MyLogin() function. Like this:
if (DBIc_IMPSET(imp_xxh)) {
...reusing imp_data...
}
else {
...existing code...
}
That way the rest of the imp_dbh_t structure gets initialised properly
(and it's how most drivers will need to do it).
> Index: dbdimp.c
> +SV *mysql_take_imp_data(SV *h, imp_xxh_t *imp_xxh, void *foo)
> +{
> + D_imp_dbh(h);
> + DBIc_ACTIVE_off(imp_dbh);
> + if (dbis->debug >= 2)
> + PerlIO_printf(DBILOGFP, "&imp_dbh->mysql: %lx\n",
> + (long) &imp_dbh->mysql);
> + return &PL_sv_yes;
> +}
The DBIc_ACTIVE_off isn't needed. In fact, looking at DBD::Mysql's
struct imp_dbh_st, mysql_take_imp_data isn't needed at all.
There's no 'preparation' the driver needs to do.
Tim.