Here's a way to write join clauses so they are more compact, more readable, and less confusing.

Many users may be unaware of the fact that there is a great tool for automatically testing MySQL called mysql-test-run. The most recent version of this program is written in perl and resides in the mysql-test directory of the MySQL source code tree.

Tests are written as scripts (.test files) which contains SQL statements and a number of meta-commands. Each .test file has a corresponding .result file which is used to verify the test output. A text-based "diff" is used to confirm that the output generated by the test is identical to the expected result.

To run a test mysql-test-run automatically sets up a clean environment and starts the MySQL server. It then executes the test scripts specified (by default all tests are executed) and verifies the output.

Most of the tests shipped with MySQL are intended to run with the MyISAM storage engine. There are also tests specifically created for other MySQL …

In MySQL Cluster a statement like

update t1 set counter=counter+1 where pk=1

can be executed in a few different ways with great difference in performance.

If we do it from the MySQL server then the update is implemented as:

read_ex(counter) ; //one round trip to MySQL Cluster, read with exclusive lock
counter++
update(counter); //one round trip to MySQL Cluster, update - write back


If we are using the NDBAPI (Cluster Direct API) we could implement it as above as well, but for the operation above, the NDBAPI offers a way of performing the update in just one round trip! This can be done using Interpreted Programs, which are instructions defined in your NDBAPI application and upon execute, sent to the ndbd kernel and executed there. Cool. What I will show you here is how to do an update using the interpreted program, which is called …

Few days ago I’ve started migration of some of my non-critical servers to Debian Etch (from Sarge). Just after first apt-get update && apt-get dist-upgrade, when apt has been upgraded, I noticed really strange (as for me) error: when I’ve tried to do “apt-get update” it worked fine, but there was annoying message like following:

# apt-get update
......
Fetched 5562B in 13s (421B/s)
Reading package lists... Done
W: There are no public key available for the following key IDs:
A70DAF536070D3A1
W: You may want to run apt-get update to correct these problems
#

UPDATE: Thanks to Kurt Fitzner we know, that:

There is already a mechanism to do this automatically:

$ apt-key update

This will obtain the necesary keys and import them. No need to go through gpg directly.

After not so …

Jon Stokes writes about this on Arstechnica, looking at a forthcoming Virginia Law Review paper entitled "The Piracy Paradox: Innovation and Intellectual Property in Fashion Design," in which two law professors investigate how the fashion industry manages to thrive despite rampant copying of clothing designs.

While it's not simply applicable to other areas (such as software), quite a few interesting observations are made - the most interesting (to me) is that making a generalistic presumption about needing very strong IP protection in order to drive innovation is just plain wrong, as the fashion industry proves. It may apply to other areas, but it's clearly not a universal rule.

I actually think that fashion and software have a lot in common, particularly in terms of the …

Continuent Sequoia, an open source database clustering solution, is part of a suite of Continuent.org open source projects that deliver high availability for virtually any database environment, including MySQL and its all engines (MyISAM, InnoBD, SolidDB, etc.).

So I've expanded the scope of the NDB/Python wrappers I was working on. Now I've got Python, Perl and C# wrappers working, at least for basic functionality. I've setup a trac instance and put a roadmap and all of that type of stuff up. If you are interested in hacking, let me know and we can talk about subversion access and all that.

For the moment, I've turned off code downloading. I'll post again when I've enabled it again.

One of the things about working for an Open Source company, is that I
feel the "need to be careful" about what post if the post involves a
competitor (or even a competing project). The problem of course is
that I deal with many open source communities that are not related to
MySQL at all (or my work on MySQL), so I do notice trends that have
nothing to do with databases.

I need a secret identity to post comments about MySQL competitors :)

Ah, screw it!

Last week I went to go buy a new 1u for project from ABMX (http://www.abmx.com/). ABMX is my current favorite white box vendor. I've
never had an issue with their computers and they are priced right.

What do I notice in the checkout box?

The linux that comes default is "Oracle Enterprise Linux 4.4".

My Link: …

A new poll we’re running on the MySQL web site shows that a lot of folks still aren’t familiar with the strict mode/sql mode setting that was introduced with MySQL 5.0. The sql_mode parameter allows you to have MySQL better enforce data integrity in terms of what data is accepted/rejected by the server. Not having this capability produced a number of gotchas in previous MySQL versions, but in 5.0, these all go away.

The sql_mode config parameter is very flexible in terms of how strong/lax you want to be with respect to having MySQL enforce data integrity. For more information on all the options available to you with the sql_mode parm, see the MySQL Manual. And for a quick primer on sql_mode and how to use it, you can …

A kind soul has contributed a Debian/Ubuntu package for the innotop MySQL and InnoDB monitor. Thanks Sebastien Estienne!