DBI::PurePerl and DBD::Oracle...

Hi all,

Just trying to understand the instructions on both these packages. Each
set of instructions explains that we do the following steps in the
following order:
- perl Makefile.PL
- make=20
- make test
- make install

Thing is, everything works fine except the make test. On the initial
pass, if I run "make test" BEFORE I run "make install" about half the
tests (50%) fail. If I then run "make install" and then RERUN "make
test" I get a 95%+ success rate. =20

I guess I'm trying to understand "why" we run the make test prior to
running the make install?

thx in advance,


Adym Lincoln
I/S Corporate - I/S Internal Applications
603-245-3744
Ext : 53744
adym.lincoln@libertymutual.com

....
It's a bug planet!!! It's an ugly planet!!! Ever feel like your
software project is going in the wrong direction.
....
=20

Re: DBI::PurePerl and DBD::Oracle...

When that many test fail with DBD::Oracle the most likely cause is that you
do not have a correct ORACLE_USERID enviornment variable set.

Chech the readmes for this. they should answer your questions

Cheers
John Scoles
----- Original Message -----
From: "Lincoln, Adym"
To:
Sent: Tuesday, August 29, 2006 8:45 AM
Subject: DBI::PurePerl and DBD::Oracle...


Hi all,

Just trying to understand the instructions on both these packages. Each
set of instructions explains that we do the following steps in the
following order:
- perl Makefile.PL
- make
- make test
- make install

Thing is, everything works fine except the make test. On the initial
pass, if I run "make test" BEFORE I run "make install" about half the
tests (50%) fail. If I then run "make install" and then RERUN "make
test" I get a 95%+ success rate.

I guess I'm trying to understand "why" we run the make test prior to
running the make install?

thx in advance,


Adym Lincoln
I/S Corporate - I/S Internal Applications
603-245-3744
Ext : 53744
adym.lincoln@libertymutual.com

....
It's a bug planet!!! It's an ugly planet!!! Ever feel like your
software project is going in the wrong direction.
....

Re: DBI::PurePerl and DBD::Oracle...

Lincoln, Adym wrote:
> Hi all,
>
> Just trying to understand the instructions on both these packages.

The main thing to understand is that DBI::PurePerl is meant to work only
with DBDs that are pure Perl (have no C/XS component) and AFAIK, that
does *not* include DBD::Oracle. You might be able to access Oracle
using DBI::PurePerl via DBD::Proxy, but I doubt it will work directly.
Is there a reason you are trying PurePerl rather than plain DBI?
> I guess I'm trying to understand "why" we run the make test prior to
> running the make install?
>
The general idea is that you should only install software that works so
you test first, then if it works, install it. In some cases you can
force an install even if the tests fail but that is opening yourself to
the possibility of failure in production.

In cases such as yours where the tests work better after install, the
chances are that there are some environment variables that are not set
in your test environment so I'd recommend looking at and setting the
environment variables required for DBD::Oracle and then doing the make test.

--
Jeff

RE: DBI::PurePerl and DBD::Oracle...

Check.

Correct. The DBI::PurePerl does not include the DBD::Oracle. They do
however work together though, as the tests conducted "after" the make
install worked against my Oracle::XE database.=20

All the failed test were from the same type of problem. The test could
not load a given DBI and/or DBD::Oracle module in the current @INC.
This kinda hinted to me that it simply couldn't find a needed *.pm file,
which might explain that I haven't set some crucial ${ENV} before
conducting the make test. Given the above @INC issue, I just figured if
I installed the modules, then the @INC issue would disappear...turns out
it worked in my case.

The reason I went with PurePerl is availability of the target machine.
We have little to no control over the Unix/AIX machines we are putting
these modules on. As such, since Perl is already on the machines, the
PurePerl distos involve less red-tape to get them in place. Also,
performance is not a concern, as we are using DBI and DBD mostly for
simple tools and reports running under cron.

Anyway, I'll play around with the install and see if any problems crop
up...


thx,

Adym Lincoln
I/S Corporate - I/S Internal Applications
603-245-3744
Ext : 53744
adym.lincoln@libertymutual.com

....
Basic Math : Music * Audio =3D Emotion
Adym Math : Green Day * JBL =3D Happiness ...
....
=20
-----Original Message-----
From: Jeff Zucker [mailto:jeff@vpservices.com]=20
Sent: Tuesday, August 29, 2006 12:49 PM
To: Lincoln, Adym
Cc: dbi-users@perl.org
Subject: Re: DBI::PurePerl and DBD::Oracle...

Lincoln, Adym wrote:
> Hi all,
>
> Just trying to understand the instructions on both these packages. =20

The main thing to understand is that DBI::PurePerl is meant to work only

with DBDs that are pure Perl (have no C/XS component) and AFAIK, that=20
does *not* include DBD::Oracle. You might be able to access Oracle=20
using DBI::PurePerl via DBD::Proxy, but I doubt it will work directly.

Is there a reason you are trying PurePerl rather than plain DBI?
> I guess I'm trying to understand "why" we run the make test prior to
> running the make install?
> =20
The general idea is that you should only install software that works so=20
you test first, then if it works, install it. In some cases you can=20
force an install even if the tests fail but that is opening yourself to=20
the possibility of failure in production.

In cases such as yours where the tests work better after install, the=20
chances are that there are some environment variables that are not set=20
in your test environment so I'd recommend looking at and setting the=20
environment variables required for DBD::Oracle and then doing the make
test.

--=20
Jeff