On Thu, Jan 08, 2004 at 04:10:33PM +0100, H.Merijn Brand wrote:
> On Thu 08 Jan 2004 11:12, Tim Bunce <[EMAIL PROTECTED]> wrote:
> > http://homepage.eircom.net/~timbunce/DBI-1.40-rc1-20040107.tar.gz
> >
> > I'd appreciate some testing on a variety of platforms, perls and drivers.
> > Thanks.
> >
> > Tim.
>
> You little rascal! I saw DBI-1.40 being uploaded while I was
> still building up this test report :)
Ever the optimist... I figured if I'd had reports that it was
okay on Windows then it was bound to be okay on unix :)
> I've finished it anyway with every combo I could use, but not
> all machines are equally fast, which you can see in the reports
>
> Enjoy, Have FUN!
>
> > =head2 Changes in DBI 1.40, 7th January 2004
> >
> > Fixed handling of CachedKids when DESTROYing threaded handles.
> > Fixed sql_user_name() in DBI::DBD::Metadata (used by write_getinfo_pm)
> > to use $dbh->{Username}. Driver authors please update your code.
>
> Where, what and how?
If you've used DBI::DBD::Metadata's write_getinfo_pm() to generate the
code for your drivers get_info() function then you should edit the
generated sql_user_name() code to use $dbh->{Username} instead of $dbh->{USER}.
> All builds automatically took the correct C compiler!
> No special environments were set
>
> Tested configurations:
> - HP-UX 11.11 perl-5.8.0/dor HP C-ANSI-C
> - HP-UX 11.11 perl-5.8.2/dor GNU gcc-3.3.2
> - HP-UX 11.11 perl-5.8.2/dor/64all GNU gcc-3.3.2/64
> - HP-UX 11.00 perl-5.6.1 HP C-ANSI-C
> - HP-UX 11.00 perl-5.8.0/dor HP C-ANSI-C
> - HP-UX 11.00 perl-5.9.0 HP C-ANSI-C
> - HP-UX 11.00 perl-5.8.2/dor GNU gcc-3.3.2
> - HP-UX 11.00 perl-5.8.2/dor/64all GNU gcc-3.3.2/64
> - HP-UX 10.20 perl-5.6.1 HP C-ANSI-C
> - HP-UX 10.20 perl-5.8.0/dor HP C-ANSI-C
> - HP-UX 10.20 perl-5.9.0 HP C-ANSI-C
> - HP-UX 10.20 perl-5.8.2/dor GNU gcc-3.3.2
> - AIX 4.3.3.0 perl-5.6.1 IBM vac-5
> - AIX 4.3.3.0 perl-5.8.0/dor IBM vac-5
> - AIX 4.3.3.0 perl-5.9.0 IBM vac-5
> - Linux 2.4.20 (SuSE 8.2) perl-5.8.1/dor/64int GNU gcc-3.3.1
> - Linux 2.4.20 (SuSE 8.2) perl-5.9.0/64int GNU gcc-3.3.2
> - Linux 2.4.20 (SuSE 8.2) perl-5.8.0/threads GNU gcc-3.3 20030226 (prerelease)
> - Cygwin-1.5.5 perl-5.8.2/threads/64int GNU gcc-3.3.1 (cygming special)
Oooh. Very nice. I'll try to wait for these reports in future!
Out of idle curiosity could you apply this patch:
--- test.pl 2003/02/26 17:56:01 11.6
+++ test.pl 2004/01/08 17:03:58
@@ -104,4 +104,4 @@
my $dur = $td->cpu_a || (1/$count); # fudge if cpu_a==0
- printf "$count NullP statement handles cycled in %.1f cpu+sys seconds (%d per
sec)\n\n",
- $dur, $count / $dur;
+ printf "%d NullP sth cycled per second on this perl $] %s (%d in %.1f cpu+sys
secs)\n\n",
+ $count/$dur, $Config{archname}, $count, $dur;
and then regenerate the test log and post the lines matching /cycled in/
and perhaps sort -n as well.
:-)
Tim.
