From: Paul DuBois
Date: April 14 2011 3:51pm
Subject: Re: bzr commit into mysql-5.1 branch (sergey.glukhov:3658) Bug#11756242
List-Archive: http://lists.mysql.com/commits/135463
Message-Id: <61FB9EF2-3DFB-4050-A578-D08B20510A8D@oracle.com>
MIME-Version: 1.0 (Apple Message framework v1084)
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

[adding innodb_dev_ww@stripped because I think the topic deserves =
wider consideration]

On Apr 14, 2011, at 4:55 AM, =D8ystein Gr=F8vlen wrote:

> On 04/14/11 10:54 AM, Sergey Glukhov wrote:
>> On 04/14/2011 12:38 PM, Bjorn Munch wrote:
>>> On 14/04 08.11, Sergey Glukhov wrote:
>>>>=20
>>>> =3D=3D=3D modified file 'mysql-test/t/analyse.test'
>>>> --- a/mysql-test/t/analyse.test 2011-03-14 18:03:22 +0000
>>>> +++ b/mysql-test/t/analyse.test 2011-04-14 08:11:57 +0000
>>>> @@ -1,6 +1,7 @@
>>>> #
>>>> # Test of procedure analyse
>>>> #
>>>> +-- source include/have_innodb.inc
>>>=20
>>> But this means the rest of the existing analyse.test will no longer =
be
>>> run without innodb. Maybe not a big deal (and not a problem for =
5.5),
>>> but I still think it's bad to disable an existing test for myisam in
>>> this way.
>> imho it' a big problem,
>> we already have 54 tests with
>> 'have_innodb' in main suite.
>=20
> Many of these tests only contain test cases for InnoDB relevant bugs, =
while general test cases have been put into a more general test.  Here =
you are limiting a general tests to when InnoDB is available.
>=20
> However, I still does not think this is a big issue since starting =
with 5.5, InnoDB shouild always be present, so the test will always be =
run. Maybe we could just drop the have_innodb inclusion going forward?


InnoDB is the default storage engine now, but does that mean it's always =
present?

Aside from the test suite issues, I think we have bigger problems. It's =
possible to build without InnoDB, and the --skip-innodb option is still =
supported. I'm unable to figure out how to get the server to start, =
though.

If InnoDB is really required, it should not be possible to build without =
it, and the --innodb/--skip-innodb options should be depecated in 5.5 =
(and made to do nothing but log a warning), and removed in a later =
release.

Also, we still have the --ignore-builtin-innodb option (and system =
variable), which really apply to the InnoDB Plugin from 5.1. What is =
supposed to happen if this option is given in 5.5? Anything? Something? =
Should it be deprecated? I don't understand its semantics now.

--=20
Paul DuBois
Oracle Corporation / MySQL Documentation Team
Madison, Wisconsin, USA
www.mysql.com