Re: Problem with rownum

On Thu, 13 Jan 2005 16:07:48 -0800, DA Morgan <damorgan_at_x.washington.edu> wrote:

>Andy Hassall wrote:
>> On Thu, 13 Jan 2005 14:27:04 -0800, DA Morgan <damorgan_at_x.washington.edu>
>> wrote:
>>>Graeme D wrote:
>>>
>>>>I don't understand why oracle doesn't have something like LIMIT x,x liek
>>>>there is in MySQL... any ideas?
>>>
>>>I don't understand why MySQL doesn't have full relational integrity and
>>>the ability to perform transactional recovery ... any ideas?
>>
>> Yes it does, and yes it can. Perhaps you're still only considering the
>> features of older versions, or haven't read the manual recently?
>>
>> A more helpful reply would be to point out how rownum in Oracle is determined,
>> i.e. it's done before a sort, so an equivalent to MySQL's limit would involve a
>> sorted subquery.
>
>Are you using the obfuscation toolkit? How in MySQL can you recover a
>corrupt database at the transaction level? I may be behind by the latest
>release but not more than that.

 There's the binary log, which looks remarkably similar to a somewhat primitive version of Oracle redo/archive logs:

 http://dev.mysql.com/doc/mysql/en/Binary_log.html

 There's then a description of rolling forwards using the binary log:

 http://dev.mysql.com/doc/mysql/en/Backing_up.html

 And on the other point, referential integrity has been supported since 3.2 - 4.1 is now the current release.

 MySQL certainly used to deserve massive amounts of scorn poured on it due to lack of even basic transactions, let alone foreign keys, but it appears to have progressed far enough to be taken seriously for certain circumstances. It's even got Oracle-style multi-versioned non-locking consistent-reads now.

 It still has various severe flaws of course, and given the choice I stick by Oracle, but they've been taking fairly big bites out of some of the main objections.

-- 
Andy Hassall / <andy@andyh.co.uk> / <http://www.andyh.co.uk>
<http://www.andyhsoftware.co.uk/space> Space: disk usage analysis tool