A working MySQL Workbench Under Linux

I must admit I’d given up trying to get MySQL Workbench working under Linux. I guess I’d spent at least 4 or 5 days full time at it, and it was just out of my league, with GTK and C++ errors. It had seemed like a loosing battle, I’ve had 3 detailed documented goes at 1.0.0. Last Post 19th Jan, and 2 with 1.0.1 Last Post 31st Jan.

Anyway, the good news is I now have MySQL Worbench working under Linux. I’ve been working with MySQL AB on a number of specific problems and you should expect a release from MySQL AB soon to address these. Here is a summary of my experience.

  • 1.0.1 removes the requirements for java dependancies
  • 1.0.1 provides a README.linux with configuration requirements, and a subset list of minimum library requirements (which needs updating)
  • libsigc++2.0.11 (an older version) is necessary, most recent versions up to .17 do not work
  • Solving the library dependancies from true source .tar.gz was difficult to manage, however I finally got to use rpm libraries to provide some consistent reproducible steps. I’ll document later as there we a number of issues with achieving this under CentOS 4.2
  • Solving the GLIBCXX_3.4.4, GLIBCXX_3.4.5, GLIBCXX_3.4.6 error. This is what finally stopped me. Thanks to my new Guru Friend Alfredo who put me on the path to solving this.
  • Underlying glXGetProcAddress() compile error required code changes (which are being incorporated in next release). This finally lead to a binary being built.
  • Executing Binary lasted a split second Segmentation fault $PRG-bin $* Debugging determined the problem to be a specific NVIDIA issue, reverting back to generic Video Drivers only left me with a OpenGL and minimum 32MB 3D accelerated graphics card error.
  • Patched code to this problem (to be incorporated in next release) has lead to a working product, at least enough to play around. Now the real fun begins.

Got heaps more to write about, however that would have slowed down the good news. More to come soon.

My thanks to the team at MySQL AB. I have received excellent care and response from the top down. I’d like to say that all the pain has been worth it in the end, and had this not been a open source company, I would have given up a long time ago.

Google might and the loss of power to the underdogs

In doing some Internet Searching I came across an interesting feature on the IBM DeveloperWorks Web Site. For the purposes of this demonstration, I’ve done some more testing to make it easier to demonstrate.

DeveloperWorks in response to a Google Search, prompts the user within the page to continue searching the IBM site with the same search criteria entered from the Google Search. In theory this would not be complicated, this referrer information is available and used in most Web Stats programs that troll log files and then provide a Section showing Search Engine Terms that directed traffic to your site.

What I surprised about, is that IBM (A large company, and key open-source proponent), restricts this functionality to Google Only. Check this out.

For each Search engine in our test, we are going to go to the respective search page, and enter the term IBM Federated Database Technology.

As a result, most search engines should provide a reference to the article at http://www-128.ibm.com/developerworks/db2/library/techarticle/0203haas/0203haas.html in the top results. We are then going to view this webpage from selecting the search results link.

The Search Engines I used were: Google · Altavista · Yahoo · Ask Jeeves · Excite



The code in effect is very simple, just some javascript.


if (document.referrer&&document.referrer!="") {
   // document.write(document.referrer);
   var q = document.referrer;
   var engine = q;
   var isG = engine.search(/google.com/i);
   var searchTerms;
   //var searchTermsForDisplay;
   if (isG != -1) {
	   var i = q.search(/q=/);
	   var q2 = q.substring(i+2);
	   var j = q2.search(/&/);
	   j = (j == -1)?q2.length:j;
	   searchTerms = q.substring(i+2,i+2+j);
	   if (searchTerms.length != 0) {
	       searchQuery(searchTerms);
	       document. write("<div id="contents"></div>");
	   }
   }
}

Federated Syntax

I’ve never used Federated. I’m waiting for the JDBC version capabilities so I can connect to a non MySQL Server (specifically Oracle). In reading the docs, I see that the syntax includes a CONNECTION String.


CREATE TABLE federated_table (
id INT(20) NOT NULL AUTO_INCREMENT,
name VARCHAR(32) NOT NULL DEFAULT '',
other INT(20) NOT NULL DEFAULT '0',
PRIMARY KEY (id),
INDEX name (name),
INDEX other_key (other)
)
ENGINE=FEDERATED
DEFAULT CHARSET=latin1
CONNECTION='mysql://root@remote_host:9306/federated/test_table';

I’m surprised in the state of the syntax for two reasons.

  • First, there are hardcoded variables in the string, this sort of breaks the rules of seperating syntax from authentication specifics
  • Second, if you have multiple FEDERATED tables, you incur a level of duplication of this information, and then multiple necessary changes if the connection string is modified

I can’t ask I’d like to see the following, as I don’t use this syntax, but I’d like to suggest a level of abstraction of this information. Could the CONNECTION point to a referenced object (e.g. CONNECTION=’HOST1′). This would solve my second point. Now, the requirements of the HOST1 object needs to contain a connection string, and where would this be stored. I don’t have an answer for that one. Perhaps somebody else could make a suggestion?

Updated
Thanks Mike for your comment, click on comments below to read. Boy it’s been been a long time since I read any ISO Standards documents. Just to clarify Mike’s point for those investigative types. SQL/MED is Management of External Data, and the official ISO document is I believe:

ISO/IEC 9075-9:2003 Information technology – Database Languages – SQL – Part 9: Management of External Data (SQL/MED) ISO Reference

Abstract
ISO/IEC 9075 defines the SQL language. The scope of the SQL language is the definition of data structure and the operations on data stored in that structure. Parts 1, 2 and 11 encompass the minimum requirements of the language. Others parts define extensions.
ISO/IEC 9075-9:2003 defines extensions to SQL to support management of external data through the use of foreign-data wrappers and datalink types.

This document makes reference to DATALINK, a syntax I remember vaguely from Oracle Days. Will need to vett that. What’s also of note in this document is the reference to using the TABLE TYPE of ‘FOREIGN’.

I wonder why MySQL went with FEDERATED as the TABLE TYPE. I’m sure this is a good reason?

Brisbane MySQL Users Group Meeting with Brian Aker

We had the privilege of Brian Aker Director of Architecture for MySQL speaking at the Brisbane MySQL Users Group this week (28 th Jan 2006). After the initial discussions on various topics, Brian got into his discussion on MySQL 5.1. I was surprised that only 2 people (myself being one of them) had 5.1 installed and being used in any way.

Here were some of the highlights of the talk. Should have taken some more notes.

MySQL 5.1 Features

Partitioning Will include Range, Hash, List and Key handling and Subpartitioning. Documentation · Forum.
Partitioning is a table structure syntax. No changes to any SQL statements are required to use partitioning within your application.

Log Tables Log files can be stored within the database. Error log will definitely not be available. Official information is sketchy.
The 5.1.6 Release notes indicate mysqld -both-log-formats and –old-log-format arguments (as of 5.1.6), but nothing yet in the The MySQL Log Files Documentation

Fast Alter Table To speed up Alter Table for example extending a VARCHAR column, only the tables .frm is rewritten, and no adjustment to table rows is required.

Replication Documentation

  • Row Based Replication (RBR) (5.1.5) Documentation
  • mysqld arguments –binlog-format and –binlog-row-event-max-size

  • Support for Cluster (NDB)
  • GUID’s support

There was a good discussion on Master/Master including comparision to PostgreSQL. This could be achieved now using triggers and federated tables, or UDF’s. Planned in 5.2 is a Conflict Resolution implementation via use of Stored Procedures, which will be a pre-requisite before any MySQL multi-master implementation.

Federated Documentation · Forum

  • Transaction Support (still with some rollback issues) Note: Documentation still lists this as a limitation
  • 5.2 will provide full XA support

XML Parsing Support (5.1.5) Article
Provides support for XPath expressions in XML content stored in a database column.

Events (5.1.6 – not yet available) Article

There was a strong discussion on occurance management of events. Obviously the ability not to have cron jobs, username/passwords etc is great, but with most scripts there is always an amount of locking management (either via filesystem or database). I think we may have convinced Brian to consider a event syntax extension that would ensure that only one instance of a given event can run at any one time. As events are seperate threads it’s important that for frequently run events, there is appropiate management to ensure re-occuring events don’t back up. My opinion is for a syntax to run only one instance of the event (e.g. SINGLE INSTANCE, SINGLE OCCURANCE) We could have recommended BRISBANE as the syntax extension, however I doubt it would be considered standard.

Also discussed was appropiate TimeZone management, particularly to ensure events scheduled appropiately across databases managed globally.

Loadable Engines
This feature may be deferred to 5.2 to have a full feature. The goal of Loadable engines allows for no need to recompile MySQL from source, of even restart an instance for any change to the engine. I’m sure internally the one reason why this is important is to allow ease of testing when writing a new storage engine. Documentation – Writing a Custom Storage Engine · Article

Cluster improvements (NDB) Documentation · RoadMap

  • replication works with cluster
  • storage on disk based records, not limited to RAM
  • indexes still remain in RAM

Benchmarking Tool mysqlslap (5.1.4) Documentation.
The list of options looks impressive, and since my 5.1 version is 5.1.4 I’ll be playing some more here.

Archive Storage Engine Documentation · Forum

  • auto increment support (5.1.6)
  • Blob Speed Improvements

Instance Manager
Available in 5.0, but being better published in 5.1 is an Instance Manager allowing change of configuration/startup/shutdown etc. Documentation

Distribution Names
As part of 5.1, the use of -standard and -max versions will be eliminated. Brian mentioned he had lost count answering the question of the difference between -max and maxDB. As at 5.1.4 (what I have) the filenames still include -max, but we were assured this would be dropped.

INFORMATION_SCHEMA
Of course as part of these changes, there will be a number of new tables in the INFORMATION_SCHEMA including ENGINES (5.1.5), PLUGINS (5.1.5), PARTITIONS (5.1.6), EVENTS (5.1.6).
Of note, there is a FILES (5.1.6) for the storage of table information?

Other References
What’s New in MySQL 5.1 MySQL Change History for Version 5.1

Future Directions for MySQL

  • This year will also be a GUI Enterprise Manager for the community edition
  • FULLTEXT for other storage engines
  • More Transaction Engines
  • Federate Tables to non MySQL tables via JDBC
  • FULLTEXT loadable parser written in C
  • Alternative stored procedure languages
  • Replication Conflict Resolution Management by user definable stored procedures
  • myforge – A soureforge/freshmeat like open source system for MySQL projects

Other Discussions

There were a number of post talk discussions. Those of interest to me included running other languages in stored procedures (myPerl/myPhp user contributions available at Brian’s open source site www.tangent.org), increasing exposure to ISP’s to offer hosting with MySQL 5.0, including best practice guidelines, and of most discussion Asterisk The Open Source PBX.

PAE Support (Part 2)

I’ve solved the PAE Support problem with my current kernel of 2.6.9-22.0.1.ELsmp. Referring to my earlier post PAE Support with the inability to get Parallels VM Software working under CentOS 4.2.

The help was thanks to a prompt response of a support request from the company (so that’s a good sign). Ultimately the trick was to boot the 2.6.9-22.0.1.EL (i.e. not SMP) kernel. Of course the kernel I’m running is the default on my laptop. So really having no idea of the actual impact (obviously something to do with the HT technology I assume).

So, problem solved, will it have an impact on system performance with this kernel? Not sure. I’ve also been told that Parallels 2.1 will have PAE support, just waiting for the response to know if buying 2.0 gives me a free upgrade to 2.1.

Solving the GLIBCXX_3.4.4, GLIBCXX_3.4.5, GLIBCXX_3.4.6 error

Let’s review the problem. I’ve got this on a number of occasions and different libraries. Here are some typical error conditions.

./mysql-workbench-bin: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.5' not found (required by ./mysql-workbench-bin)
Error: Missing Dependency: libstdc++.so.6(GLIBCXX_3.4.6) is needed by package glibmm24

Special thanks to my new Guru friend Alfredo who put me on the path.

$ su -
$ cd /src/rpm
$ wget ftp://rpmfind.net/linux/fedora/extras/4/SRPMS/libsigc++20-2.0.11-1.src.rpm
$ rpmbuild --rebuild libsigc++20-2.0.11-1.src.rpm
$ cd /usr/src/redhat/RPMS/i386/
$ rpm -Uvh libsigc++20*.rpm
Preparing...                ########################################### [100%]
   1:libsigc++20            ########################################### [ 50%]
   2:libsigc++20-devel      ########################################### [100%]

Woot! It was that simple, download the rpm source, compile and install.

What does this ultimately mean? Well, something in your OS configuration isn’t the same as the platform used to build the rpm’s. In my case, I’m running CentOS 4.2, which was been upgraded from CentOS 4.0 via the Network Notification Tool. I’m also trying to install a Fedora FC4 rpm, on a RHEL platform. Anyway, it’s a simple solution and it works and that’s all we need for those that get stuck with this problem.

A detailed History of the problem

$ su -
$ yum install libsigc++20
Setting up Install Process
Setting up repositories
addons                    100% |=========================|  951 B    00:00
update                    100% |=========================|  951 B    00:00
nr-production             100% |=========================| 1.0 kB    00:00
base                      100% |=========================| 1.1 kB    00:00
nr-testing                100% |=========================| 1.0 kB    00:00
nr-mono                   100% |=========================| 1.0 kB    00:00
freshrpms                 100% |=========================|  951 B    00:00
extras                    100% |=========================| 1.1 kB    00:00
Reading repository metadata in from local files
primary.xml.gz            100% |=========================|  35 kB    00:01
nr-mono   : ################################################## 143/143
Added 131 new packages, deleted 87 old in 1.31 seconds
Parsing package install arguments
Resolving Dependencies
--> Populating transaction set with selected packages. Please wait.
---> Package libsigc++20.i386 0:2.0.17-1.1.fc4.nr set to be updated
--> Running transaction check
--> Processing Dependency: libstdc++.so.6(GLIBCXX_3.4.4) for package: libsigc++20
--> Processing Dependency: libstdc++.so.6(GLIBCXX_3.4.6) for package: libsigc++20
--> Finished Dependency Resolution
Error: Missing Dependency: libstdc++.so.6(GLIBCXX_3.4.4) is needed by package libsigc++20
Error: Missing Dependency: libstdc++.so.6(GLIBCXX_3.4.6) is needed by package libsigc++20

So doing some investigation.

$ yum info libstdc++
Setting up repositories
addons                    100% |=========================|  951 B    00:00
update                    100% |=========================|  951 B    00:00
nr-production             100% |=========================| 1.0 kB    00:00
base                      100% |=========================| 1.1 kB    00:00
nr-testing                100% |=========================| 1.0 kB    00:00
nr-mono                   100% |=========================| 1.0 kB    00:00
freshrpms                 100% |=========================|  951 B    00:00
extras                    100% |=========================| 1.1 kB    00:00
Reading repository metadata in from local files
Installed Packages
Name   : libstdc++
Arch   : i386
Version: 3.4.4
Release: 2
Size   : 774 k
Repo   : installed
Summary: GNU Standard C++ Library

Description:
 The libstdc++ package contains a rewritten standard compliant GCC Standard
C++ Library.


$ rpm -ql libstdc++
/usr/lib/libstdc++.so.6
/usr/lib/libstdc++.so.6.0.3


$ rpm -qaR libstdc++
/sbin/ldconfig
/sbin/ldconfig
libc.so.6
libc.so.6(GLIBC_2.0)
libc.so.6(GLIBC_2.1)
libc.so.6(GLIBC_2.1.3)
libc.so.6(GLIBC_2.2)
libc.so.6(GLIBC_2.3)
libgcc_s.so.1
libgcc_s.so.1(GCC_3.0)
libgcc_s.so.1(GCC_3.3)
libgcc_s.so.1(GLIBC_2.0)
libm.so.6
rpmlib(CompressedFileNames) < = 3.0.4-1
rpmlib(PayloadFilesHavePrefix) <= 4.0-1
rpmlib(VersionedDependencies) <= 3.0.3-1

Burning DVD's under CentOS 4.2

Attempt to update XCDRoast. Instructions at http://www.xcdroast.org/manual/dvd.html


su -
/usr/lib/xcdroast-0.98/bin
wget ftp://ftp.berlios.de/pub/cdrecord/ProDVD/cdrecord-prodvd-2.01a12-i586-pc-linux-gnu
mv cdrecord-prodvd-2.01a12-i586-pc-linux-gnu cdrecord.prodvd
chmod 755 cdrecord.prodvd
exit
# xcdroast not configured for root usage
xcdroast &
# You will get a warning message, this confirms that prodvd is installed

SetUp | Options | CDR_SECURITY_KEY
Now get 4.7GB option, but when attempting to write kept getting can’t init drive option, then it would eject DVD?

So I’ve moved to trying GnomeBaker


su -
cd /src/rpm
wget http://repo.nrpms.net/gnomebaker/0.5.0/RPMS/gnomebaker-0.5.0-1.1.fc4.nr.i386.rpm
rpm -ivh gnomebaker-0.5.0-1.1.fc4.nr.i386.rpm

There were some dependancies, but I’ve not noted these here.

Another Enterprise LAMP stack provider

ActiveGrid, the Enterprise LAMP company, provides a service-oriented application platform built on the lightweight architecture of the proven LAMP software infrastructure stack. ActiveGrid Enterprise LAMP simplifies and speeds the development of service-oriented applications that weave together existing enterprise systems into new rich web applications and services. ActiveGrid Enterprise LAMP applications can be flexibly deployed on grids of commodity machines or at virtually any ISP.

Please see http://www.activegrid.com for more information.

Building MySQL Workbench 1.0.1 for Linux (Part 2)

Following my earlier post of MySQL Workbench 1.0.1 for Linux and logging a MySQL Bug, I’ve had the Bug verified, and the a further update of a compiler success. Details of compile from Bug #16880

Compiled with:

- glibmm-2.8.1
- gtk+-2.8.8
- libsigc++-2.0.11

I’ve got:


libsigc++-2.0.17
glib-2.6.6
glibmm-2.6.1
atk-1.9.0
pango-1.8.2
gtk+-2.6.9
gtkmm-2.6.5
lua-5.0.2

So starting with the obvious downgrading of libsigc++


$ su -
$ cd /src
$ wget http://ftp.gnome.org/pub/GNOME/sources/libsigc++/2.0/libsigc++-2.0.11.tar.gz
$ tar xvfz libsigc++-2.0.11.tar.gz
$ cd libsigc++-2.0.11
$ ./configure --prefix=/usr
make
make install

Now to retrace the MySQL compiling.


$ su -
$ cd /src
$ wget ftp://ftp.mysql.com/pub/mysql/download/mysql-workbench-1.0.1.tar.gz
$ tar xvfz mysql-workbench-1.0.1.tar.gz
$ cd mysql-workbench-1.0.1
$ cd mysql-gui-common
$ ./configure --enable-grt --enable-canvas
$ make

No errors, a sound start.


$ cd ../mysql-workbench/
$ ./configure
$ make
../../../mysql-gui-common/library_gc/source/libgcanvas.a(myx_gc_gl_helper.o)(.text+0x1b5): In function `getProcAddress(char const*)':
/src/mysql-workbench-1.0.1/mysql-gui-common/library_gc/source/myx_gc_gl_helper.cpp:264: undefined reference to `glXGetProcAddress'
collect2: ld returned 1 exit status
make[3]: *** [mysql-workbench-bin] Error 1
make[3]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source/linux'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source/linux'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source'
make: *** [all-recursive] Error 1

Crash and burn. So try to reproduce with the more accurate libraries. Of course it’s a bit of trial and error, here is the end result in order, but it was much more difficult as previously to get this.


$ cd /src
$ wget ftp://ftp.gtk.org/pub/gtk/v2.8/pango-1.10.2.tar.gz
$ tar xvfz pango-1.10.2.tar.gz
$ cd pango-1.10.2
$ ./configure --prefix=/usr
$ make
$ make install
$ wget ftp://ftp.gtk.org/pub/gtk/v2.8/glib-2.8.6.tar.gz
$ tar xvfz glib-2.8.6.tar.gz
$ cd glib-2.8.6
$ ./configure --prefix=/usr
$ make
$ make lib
$ wget ftp://ftp.gtk.org/pub/gtk/v2.8/atk-1.10.3.tar.gz
$ tar xvfz atk-1.10.3.tar.gz
$ cd atk-1.10.3
$ ./configure --prefix=/usr
$ make
$ make install
# (sigc++-2.0 >= 2.0.0 glib-2.0 >= 2.8.0 gobject-2.0 >= 2.8.0 gmodule-2.0 >= 2.8.0)
# Latest is 2.8.4
$ wget http://ftp.gnome.org/pub/GNOME/sources/glibmm/2.8/glibmm-2.8.1.tar.gz
$ tar xvfz glibmm-2.8.1.tar.gz
$ cd glibmm-2.8.1
$ ./configure --prefix=/usr
$ make
$ make install
# New Requirements for gtk 2.8.8
$ wget ftp://ftp.gtk.org/pub/gtk/v2.8/dependencies/cairo-1.0.2.tar.gz
$ tar xvfz cairo-1.0.2.tar.gz
$ cd cairo-1.0.2
$ ./configure --prefix=/usr
$ make
$ make install
# Pango not found. Pango built with Cairo support is required
$ cd /src/pango-1.10.2
$ ./configure --prefix=/usr --with-cairo=YES
$ make
$ make install
$wget ftp://ftp.gtk.org/pub/gtk/v2.8/gtk+-2.8.8.tar.gz
$tar xvfz gtk+-2.8.8.tar.gz
$cd gtk+-2.8.8
# Requirements (glib-2.0 >= 2.7.1 atk >= 1.0.1 pango >= 1.9.0 cairo >= 0.9.2)
$ ./configure --prefix=/usr
$ make
$ make install
# gtkmm Latest is 2.8.2
$ wget http://ftp.gnome.org/pub/GNOME/sources/gtkmm/2.8/gtkmm-2.8.2.tar.gz
tar xvfz gtkmm-2.8.2.tar.gz
cd gtkmm-2.8.2
./configure --prefix=/usr
# Note, this takes by far the longest to compile
make
make install

SideNote: This was a maze to get the right dependancies for the right products. It took many hours.

And around the merry-go-round we go again.


cd /src/mysql-workbench-1.0.1/mysql-gui-common
$ ./configure --enable-grt --enable-canvas
$ make
$ make install
$ cd ../mysql-workbench/
$ ./configure
$ make
/usr/lib/python2.3/config/libpython2.3.a(posixmodule.o)(.text+0x3c5a): In function `posix_tmpnam':
: warning: the use of `tmpnam_r' is dangerous, better use `mkstemp'
/usr/lib/python2.3/config/libpython2.3.a(posixmodule.o)(.text+0x3b94): In function `posix_tempnam':
: warning: the use of `tempnam' is dangerous, better use `mkstemp'
../../../mysql-gui-common/library_gc/source/libgcanvas.a(myx_gc_gl_helper.o)(.text+0x1b5): In function `getProcAddress(char const*)':
/src/mysql-workbench-1.0.1/mysql-gui-common/library_gc/source/myx_gc_gl_helper.cpp:264: undefined reference to `glXGetProcAddress'
collect2: ld returned 1 exit status
make[3]: *** [mysql-workbench-bin] Error 1
make[3]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source/linux'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source/linux'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-workbench/source'
make: *** [all-recursive] Error 1

Same result, so after another 4 or 5 hours, I still have nothing.

Update
Some more research (I really should devote my time to other pursuits). An official comment from NVidia Corporation on their forums. View Thread

This is not a NVIDIA bug, glXGetProcAddress() is part of GLX 1.4, which the NVIDIA Linux/UNIX graphics driver doesn’t claim to support (nor do I think was GLX 1.4 ever finalized); you should always use glXGetProcAddressARB().

Woot!

I had a friend go woot in an IM message today, and my response was something like. Yes, haven’t been to woot.com for a while, which lead to statement “I didn’t know there was a site”.

Well yes folks, there is a www.woot.com, and it’s way cool. Of course you can go there, and unless you read more then the first page you might wonder what all the fuss is about. I’ve taken the liberty of stealing some content from they FAQ.

What is Woot?
Woot.com is an online store and community that focuses on selling cool stuff cheap. It started as an employee-store slash market-testing type of place for an electronics distributor, but it’s taken on a life of its own.
I see only one item, do you sell anything else?
No. We sell one item per day until it is sold out or until 11:59pm central time when it is replaced (see next entry for details). However, each item we sell is in stock and typically ships within 2-3 business days.
What is the schedule for new items?
The short answer: we offer a new item every single day. The details: a new product is released every morning at 12am central time, seven days a week.

You can read more at What is Woot?

So what kind of stuff do they sell. Well you can look at the product archives here. Anything from a Creative Labs USB Bluetooth Adaptor to a Western Digital Caviar 250GB Hard Drive to a InFocus Ultra-Thin 61″ DLP HDTV with Blender.

Woot!

The response to a good book review.

I read many years ago now Maverick written by Ricardo Semler. It was a brillant book,that I greatly enjoyed. This book is based on what Ricardo did when he took over his father’s company in Brazil called Semco. His book was on the top sellers list for a very long time.

So what did he do that was so great. “Semco has let his employees set the terms of their employment: hours, wages, even their office technology.” Of course one line can’t describe the book.

I recommended this book to a friend, and then I eventually bought him a copy some months ago. Well, this evening in a meeting I’d found that he actually got around to reading Maverick. He was so impressed he bought 9 copies to give to other business collegues. You can’t get a better response to a good book review then that.

If you have never heard of him, Google has some matches 90,000 matches for Ricardo Semler.

Choosing a new business IT system

I know of this not-for-profit organisation that is seeking to find the ideal solution for all of their IT needs. As with most organisations, wish lists are long, and in this case, the pockets are very shallow.

Now there is a certain amount of history, and politics (all organisations have this), so I’ll just focus on the facts at hand. A new system is sought, the budget is limited, there is no IT manager for the organisation, there are no detailed user or functional requirements for product comparision. So what do you do.

I’m not going to be involved in the process in any way, so what advice in guidelines could I give that would enable the parties to at least consider. Remembering that the Organisation has no IT manager, or internal experienced IT professional to advise. (This is of course the most dangerous position, the worst business critical IT decisions are made by non-IT people simply because they are putting the business needs first (that’s their job), but if your organisation is totally dependent on IT, or nearly totally dependent on IT (it could survive a short time, but you would need a lot more staff) how can the best views be presented without an IT expert, even totally independent from the organisation.)

Here is the quick guide.

1. There is an existing operational (but not fully functionality, obviously the reason for moving) system. What are the top 10 points of functionality that you business currently uses and gains most benefit from (also considering cost savings in money and staff resources).

2. What are the top 10 points of functionality the current system does not currently fully provide (obviously for moving). Be specific.

3. Given the case of globalisation, overseas products will be considered. The following are the Internationalisation (I18N) points the system must have (like the permenent members of the UN Security Council), if one fails the product fails regardless.

  1. Must be able to handle input and display of dates in local format
  2. Must operate in the Timezone of the business (for off-site managed options)
  3. Must provide correct currency support
  4. Must provide english (if the product is developed in a foreign langauge)
  5. Must provide correct Financial Year management
  6. Must provide correct local taxation requirements
  7. Must provide local integrations, for example to payment gateways, shipping providers etc.

So what you have from here are 7 must have points (one fails, end of evaluation), and then 20 points next on the list which would form a percentage of 100% if all 20 are met. Obviously using the existing system as a guide, the percentage would be 50%, 10 points it has, and 10 it doesn’t have. Simple.

What is an acceptable fit for the organisation, 50%, 75%, 80%. If you were purchasing an off the shelf product, I would think 80% would be a suitable go-live indicator. Not idea, but sufficient for survival. Doing this evaluation up front is a necessary step of comparing apples with apples.
No product will make 100%, the next part of the evaluation after getting these initial percentages is to look at the cost (in dollars and time) that it would take to move each product to the minimum acceptable fit, in my example 80%. Of course you may want to consider the cost to move to 95%.

Lastly, your organisation must consult in this case an external IT party. Somebody/some organisation which will look at the feature requirements (which are very limited), cost/benefit analysis and provide a best recommendation for the business, a decision made by mind, and not a decision made by heart.

So in summary.

  • Determine your minumum requirements
  • Determine the minimum acceptable percentage fit
  • Determine the cost (time and money) for each product to meet minimum acceptable percentage
  • Consult an independent person/body that can provide professional advice

Only at this point can you make a better informed decision. Will it be the right decision? Well that’s another question.

Blog/Wiki Spamming – What makes your blood boil

Well this is low. I’ve just been spammed on my Wiki. And it was cunning, I just found it by accident. An enterprising hacker embedded into my Home Page hidden links that were not visible via normal page view, but ultimately would be via a search bot or some other means.

Even better the pricks even deleted content on one page. Here is an example diff of pages. I even posted a few days ago What makes your blood boil? and that was just an article on a news site with misleading dated information. I could see the opportunity for more blood boiling articles, but it doesn’t solve the problem. The problem is, I don’t know how to solve the problem.

So what can you do other then clean up this hacker mess, and put in checks to find this, and then continually revise these checks. What bodies can you complain to about URL’s listed, how can you get these removed from the web. That’s what I really want to know.

MediaWiki, the blog software I use did provide a number of manual tools to help identify and correct the information, but only in response to my accidental discovery. I was able to review a History of my Home Page, and then I could drill down to all User Contributions. But it’s a manual process to cleanup. Needless to say I’ll be developing something to more easily identify this and provide. You can also use Block user to stop IP addresses. I’ll also be adding them to my firewall rules.

For those that use Wiki’s or Blog’s, please let me know your software used, and I’ll endevour over time to perhaps provide tips for these products.

As per a request by a MySQL user, I opened my Blog to allow unregistered user comments. I did this reluctently as I wanted that ‘confirmed opt-in’ step, of a user registration, however I had faith in the community. Well I’ve since turned that off as well. I’ve had SPAM posted to my blog, and most recently a post in which I wanted to respond to, but as the information supplied for email as clearly invalid I could not.

So the penalty is users wishing to make comments have to register, it’s quite painless, however the downside is as soon as you enter an email on a website what’s going to happen to it. And this could happen unwillingly. While the Blog owner as no intentions of using registered emails, a flaw in software could expose information for unlawful access.

I’ve implemented a solution to this, and I’m trialling another. The first is I have a virtual domain, not just a virtual email. You have all heard about getting a free email account at a HotMail or Yahoo or etc, and then for subscriptions you use this email, therefore protecting your own identity email from obvious spam. Well I go one step further. I use a whole domain (in this case myvirtualemail.biz), so it costs ~USD$10 per year, big deal, but what I do is I create a new alias for everything I subscribe to. I then have total control of where I send the email, and what I want to do with it. I can also track if I get any spam, which email alias it was addressed to, and then I have a greater granularity of source of the problem. I can just simply trash the alias, therefore no more SPAM in my inbox. The problem is, the SPAM still comes, it takes network traffic, bandwidth and CPU, it gets rejected and then takes even more network traffic and bandwidth.

Now you have me started on a completely different topic of Authenticated Mail. Email is long since been poisoned, and it should be left to die, and out of the ashes a new Phoenix . Cut out the source of SPAM by ensuring the sender is Authenticated. I guess a topic for another time, but it’s the impossibility of moving people off email that makes it enviable.

And the other trial, I’m using GMail to do my filtering for me. I’ve converted my default DNS administrator email (which gets a lot of spam as it’s public on domain registrars), to an alias that redirects to a gmail account, and then from this I forward filtered mail to a POP account. With gmails strict policy on attachments, this approach may not be ideal for the end user, but for this purpose it’s ideal. Still testing this idea, but it’s looking good so far. BTW, if you want a free gmail account, let me know and I’ll send an invitation.

If only I have and enterprising billionaire to fund some of my ideas, I’d really like to be one person that makes a difference, and my difference is to rid us of obvious stuff that destroys our time. And don’t get me started on Viruses. Simple solution, If the Windows OS was read-only, and there was some authentication process for software installation, we could probably rid the world of viruses. I must admit, the VMWare player running a virtual machine solves the problem as well.

Update

Some info on Setting user rights in MediaWiki
in WIKI_HOME/LocalSettings.php I added

$wgWhitelistEdit = true;
$wgShowIPinHeader = false; # For non-logged in users
$wgSysopUserBans = false; # Allow sysops to ban logged-in users
$wgSysopRangeBans = false; # Allow sysops to ban IP ranges

Other references: Wikipedia:Vandalism, Counter Vandalism Unit

MySQL Sakila Sample Application

I’m sure you are all aware by now of Mike Hillyer’s MySQL Sakila Sample Database that will be launched at the MySQL Conference. We now have an official MySQL Forum for this as well.

As part of leveraging this existing database, and using this for the basis of my MySQL Conference presentation on MySQL for Oracle Developers, I’ve released the first version of my MySQL Sakila Sample Application at http://sakila.arabx.com.au which I would very much like some feedback on. Please use the official MySQL Forum for any comments, suggestions or complaints. Please Note I am still very much in the planning and design phase.

We also have an Unofficial Wiki that describes a little more of the concept and purpose the Sample Application, and a call for others to get onboard to design and develop their own versions in varying languages.

So the sample application, what does this showcase? For now, work has been on the presentation of data, as we finalise the schema and data. In addition the application has been designed to be more self documenting describing via the top menu options the functions available, and business logic considerations, specific MySQL features and a schema to show the underlying tables in question. Look at Admin or Film for an initial example.

Here is a quick list of the functions of the MySQL Sakila Sample Application. For now there is not user authentication so it’s open for all to view.

  • Home
  • Customers
    • Index (NF), Search/List, Add, View (NF)
  • Rentals
    • Index(NF), New, Search (NF), Return(NF), Overdue(NF), Out(NF)
  • Film
    • Index, Movie List, Actor List, Categories, Languages
  • Reports
    • Index (NF), Top Film Rentals, Top Customer Rentals
  • Admin
    • Index, Staff, Stores, Countries, Cities

(NF) – Not Functional – Please note, as the data hasn’t been finalised, some of the data is my own patch just to display functionality.
More information on what’s available is in the Admin|Documentation page.

MySQL Workbench 1.0.1 for Linux

Just released at the MySQL Forums yesterday an updated source version of MySQL Workbench for Linux available at ftp://ftp.mysql.com/pub/mysql/download/mysql-workbench-1.0.1.tar.gz.

So can Version 1.0.1 compile when I had no success with compiling 1.0.0?


$ su -
$ cd /src
$ wget ftp://ftp.mysql.com/pub/mysql/download/mysql-workbench-1.0.1.tar.gz
$ tar xvfz mysql-workbench-1.0.1.tar.gz
$ cd mysql-workbench-1.0.1
$ cd mysql-gui-common
$ ./configure --enable-grt --enable-canvas
$ make
MySQLGRT/MGRTValueTree.cc:255: instantiated from here
/usr/include/sigc++-2.0/sigc++/adaptors/bound_argument.h:158: error: 'const class sigc::bound_argument >&>' has no member named 'visit'
make[3]: *** [MGRTValueTree.o] Error 1
make[3]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-gui-common/source/linux'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-gui-common/source'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/src/mysql-workbench-1.0.1/mysql-gui-common'
make: *** [all] Error 2

Arrrrrggggghhhhh!. What do I have to do to get this beast to compile. I am running the most current libsigc++ of 2.0.17, available for download at http://ftp.gnome.org/pub/GNOME/sources/libsigc++/2.0/

The job made easier this time by the src archive containing the all important README that provides both dependancies and required additional configuration arguments. However it only stated libsigc++ 2.0 which is what I have.

more /src/mysql-workbench-1.0.1/mysql-workbench/README.linux

Compiling Workbench from Scratch
================================

Required Software:
------------------

Much of the requirements are common for all GUI tools shipped by MySQL (such as
Administrator and Query Browser). Instead of building everything by hand, you
should try using the packages supplied by your distribution, preferably using
some tool like apt-get, smart or even yum. Do not forget that you need the -devel
(or -dev) version of each pacakge as well as the main one.

- gtkmm-2.4 or newer and it's dependencies, such as:
        - glibmm-2.4
        - libsigc++-2.0
        - gtk-2.4 (or whatever is required by gtkmm)
- libpcre-5
- MySQL client libraries and headers for MySQL >= 5.0
- OpenGL libraries and headers
- lua 5 (www.lua.org)

Building:
---------

- unpack workbench
- cd mysql-gui-common
- ./configure --enable-grt --enable-canvas
- make
- make install
- cd ../mysql-workbench
- ./configure
- make
- make install

I’ve logged as Bug #16880. I understand there are many variables in different Linux Operating Systems, and libraries, however if I meet the minimum requirements, I have no other options. Perhaps some specifics of what exact versions of required libraries were used to compile in an internal environment would be helpful. Are programs compiled on a generic OS distro and not a developers machine, so it can be reproduced and is testable.

Updated
In public response to pablo’s comment (click on comments to see, which I think is unqualified, please give more details), MySQL AB do read this, I have had positive comments from MySQL AB staff, and they have been proactive since I’ve raised this an earlier issues. In the space of some 4 hours I’ve had a confirmation and verification of my listed Bug, so in this case, response is excellent. Previously with 1.0.0 I was frustrated in a number of areas, first I’ve been waiting probably 8 months to get my hands on this product, at present I don’t have an alternative Open Source Linux Modelling tool supporting MySQL 5. Second, there seemed to be slow progress or perhaps insufficient feedback of progress from MySQL AB, but unless we raise our voice how will they know.

Again in defense of MySQL AB (of which I am not associated with), this is an developing Open Source Product and what we as the community ultimately get out of the product it what we collectively put in. If for example we just say it’s buggy, but don’t report it, seek advice, try to use our best abilities to reproduce and assist in providing a better product. Unfortunately the language and GUI requirements are skills I am not confident in. If they were I would be more involved in coding to correct this problem to move forward.

Another reason for presenting my information is to seek advice if others are experiencing these issues, to raise awareness to MySQL AB, who are an organisation with limited resources yet continue to provide what I consider is an excellent product for it’s current marketplace targets. I’m keen to see a working product because then I can provide 15 years of database modelling experience in large systems design in the review and testing of the product, and hopefully contribute in identifying useful and practical functionality to continully improve the product.

Again, speaking on MySQL AB’s behalf, the introduction of GUI products released last year on which the MySQL Workbench are built on, are only new products and new products take time to mature. Of course I’ve love a fully working product, but I’m not providing finances to millions or billions in R&D for software development.

I think it’s important that we all embrace the ideals of Open Source. Sure they are not perfect at this time, but this is not the ideal of one company to change the world of computing and computer software, but to all individuals to have an opportunity to contribute.

Pablo, I would be most happy to take this discussion offline and ask some more questions regarding the concerns you have made in your post, and try to help better understand your concerns, but without a suitable valid email address I’m unable to respond.

Downgrading a MySQL schema from 5 to 4 (Part 2)

As requested by Frank, here are the working parts of my earlier Downgrading a MySQL schema from 5 to 4 article.

The Problem

To recap, I received a MySQL Version 5.0 schema via a sql file, however I was unable to upgrade from MySQL 4.0 to MySQL 5.0 on my old RedHat 7.3 production server. As an interim solution, I still wanted the schema and data to allow for initial development (without the 5 specific features including Views,Triggers and Procedures/Functions). However the MySQL 5.0 SQL file would not run in MySQL 4.0.

Sample

Here is a small subset of the MySQL Sakila Sample Database schema to demonstrate the problem.

DROP SCHEMA IF EXISTS sakila;
CREATE SCHEMA sakila;
USE sakila;
--
-- Table structure for table `actor`
--
CREATE TABLE actor (
  actor_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
  first_name VARCHAR(45) NOT NULL,
  last_name VARCHAR(45) NOT NULL,
  last_update TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
  active BOOLEAN NOT NULL DEFAULT TRUE,
  PRIMARY KEY  (actor_id),
  KEY idx_actor_last_name (last_name)
)ENGINE=InnoDB DEFAULT CHARSET=utf8;

Conversion

The following commands produced the output valid for a MySQL 4.0 database.


$ dos2unix sakila-schema-0.2.sql
$ cat sakila-schema-0.2.sql | sed -f mysql5to4.sed >sakila-schema-0.2.mysql4

NOTE: These commands are Linux Commands. There are probably Windows compatible commands however without Windows around to test this, I’m not going to offer any advice here. Anybody with any experience here please advise.
If you don’t have Linux, then get it. Sound to hard, well start with a Live CD, for example Knoppix or Ubuntu. These will allow you to boot of CD and it will never affect your Windows machine. If you don’t have broadband or a CD Burner, look at buying a computer magazine. You will find quite regularly options or Live CD’s. Of if you can’t afford that, Ubuntu – ShipIt will ship you the CD for free.
Even better then this, you can now use VMWare Player and have a working Linux environment running in parallel with your Windows at the same time, and it will not affect Windows in any way. You just need the VMWare Player, then a suitable Virtual machine. If there is a single person with an excuse why they can’t experiment with Linux, your not embracing Open Source. Start today, start right now, I’ve given you the links.

Results

DROP DATABASE IF EXISTS sakila;
CREATE DATABASE sakila;
USE sakila;
--
-- Table structure for table `actor`
--
CREATE TABLE actor (
  actor_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
  first_name VARCHAR(45) NOT NULL,
  last_name VARCHAR(45) NOT NULL,
  last_update TIMESTAMP NOT NULL,
  active TINYINT NOT NULL DEFAULT TRUE,
  PRIMARY KEY  (actor_id),
  KEY idx_actor_last_name (last_name)
)TYPE=InnoDB;

The Code

Ops, almost forgot the important bit. Note, the syntax in this file is specific, characters and spaces are significant. My advice if you have no idea about ex regular expresssions and syntax, don’t touch.

The contents of mysql5to4.sed

s/CREATE SCHEMA/CREATE DATABASE/
s/DROP SCHEMA/DROP DATABASE/
s/ DEFAULT CHARSET=utf8;$/;/
s/ENGINE=/TYPE=/
s/ ON UPDATE CURRENT_TIMESTAMP//
s/ DEFAULT CURRENT_TIMESTAMP//
s/BOOL /TINYINT /
s/BOOLEAN /TINYINT /
/^DELIMITER/d

Further Work

  • These workarounds work on a generated mysqldump file. For example, I bank on implied standards of UPPERCASE in the syntax of commands.
  • I don’t cater for situations where DEFAULT CURRENT_TIMESTAMP is defined on a TIMESTAMP column which is not the first listed in the database table. This could be catered for with some more shell scripting with awk (More about awk 1, 2, 3)
  • I don’t cater for syntax situations not currently described in the Sakila Sample Database.
  • For now, I’ve manually removed the TRIGGERS, VIEWS, PROCEDURES and FUNCTIONS, however there would be good success via shell scripting to also archieve this.

MySQL Alternative

Morgan provided a suitable syntax for mysqldump in his Compatibility between MySQL Versions article, however I found this not to work completely successful. Here are my findings.

Syntax


mysqldump --compatible=name

Example

Using 5.1.4-alpha-max under Linux (not glib23)

$ mysqldump -P3307 -hlocalhost.localdomain -uroot -d --compatible=mysql40 sakila

The resultant file had the following lines, I’ve just cut and paste to show issues.
Using 4.0.13-standard.


mysql> DELIMITER ;;
ERROR 1064: You have an error in your SQL syntax. Check the manual that corresponds to your MySQL server version for the right syntax to use near 'DELIMITER' at line 1

Using 4.1.10a-standard

mysql> DELIMITER ;
ERROR:
DELIMITER must be followed by a 'delimiter' character or string

This invalid handling of DELIMITER has a downstream affect when it gets to triggers.

mysql> /*!50003 CREATE TRIGGER `ins_film` AFTER INSERT ON `film` FOR EACH ROW BEGIN
-> INSERT INTO film_text (film_id, title, description)
-> VALUES (new.film_id, new.title, new.description);
Query OK, 0 rows affected (0.00 sec)
mysql> END */;;
ERROR 1064: You have an error in your SQL syntax. Check the manual that corresponds to your MySQL server version for the right syntax to use near 'END */' at line 1

I Need to get around to checking the Bug Database to see if these need to be logged.

What makes your blood boil?

It’s appalling that in this day of technological advancements and communication, the excuse for publishing dated information just doesn’t fly. 50 or 100 years ago you could be excused for writing something that was 6 months out of date, yet this article “Which Database Is Right For You?” Dated (2006-01-05) states:

So what doesn’t MySQL support? Today, MySQL doesn’t support views (‘virtual’ tables made from other tables), stored procedures (small programs that can be stored in the database) or triggers (actions that the database can be told to do automatically when certain things happen). However, many of these features are promised in future versions.

Well as I finished the article, looking forward to the contact details to notify this “writer” of their dated information, I was beaten to the punch.

Editor’s Note – Correction: We were contacted by Jay Pipes,Community Relations Manager, North America MySQL, Inc. who says that “MySQL 5 (the current stable release of MySQL) does indeed support stored procedures, views, triggers, functions, and many more features.”

Why can’t they put that note in the MySQL section of the article, it can easily get missed at the bottom.

It was only recently that I went out to by a magazine on referral to read an in depth article “The Usual Suspects”. I was most disappointed in it’s dated and in someways inaccurate content. You can read my comments at Review of Database Magazine Article – “The Usual Suspects”

Sequences in MySQL

One piece of SQL functionality that doesn’t appear to have any consistency or an ANSI SQL Standard is the management of system generated sequential numbers, used for example in suggorate keys.

MySQL uses AUTO_INCREMENT which serves the purposes adequately, however in my documenting of differences with Oracle in my upcoming MySQL Conference presentation “MySQL for Oracle Developers” there a number of key differences with Oracle’s SEQUENCE usage.

MySQL AUTO_INCREMENT to Oracle SEQUENCE Differences

  • AUTO_INCREMENT is limited to one column per table
  • AUTO_INCREMENT must be assigned to a specific table.column (not allowing multi table use)
  • AUTO_INCREMENT is INSERTed as a not specified column, or a value of NULL

The MaxDB Reserved Words list includes SEQUENCE for the CREATE SEQUENCE however I’ve never used MaxDB. Other popular open source products such as PostgreSQL and Ingres use sequences. Refer to the references section for more details.

Usage

The following provides an example sytax usage within MySQL and Oracle.

MySQL

CREATE TABLE Movie(
id           INT NOT NULL AUTO_INCREMENT,
name     VARCHAR(60) NOT NULL,
released YEAR NOT NULL,
PRIMARY KEY (id)
) ENGINE=InnoDB;


INSERT INTO Movie (name,released) VALUES ('Gladiator',2000);
INSERT INTO Movie (id,name,released) VALUES (NULL,'The Bourne Identity',1998);

Oracle

CREATE TABLE Movie(
id          INT NOT NULL,
name     VARCHAR2(60) NOT NULL,
released INT NOT NULL,
PRIMARY KEY (id)
);
CREATE SEQUENCE MovieSeq;


INSERT INTO Movie (id,name,released) VALUES (MovieSeq.NEXTVAL,'Gladiator',2000);

You can within Oracle use a Before Insert trigger to simulate handling of the MySQL Insert syntax. Note: Within Oracle you will require a SEQUENCE per table and a TRIGGER per table. Oracle supports multiple triggers of the same type per table (not sure if MySQL supports this).

CREATE OR REPLACE TRIGGER BRI_MOVIE_TRG
BEFORE INSERT ON Movie
FOR EACH ROW
BEGIN
  SELECT MovieSeq.NEXTVAL INTO :new.id FROM DUAL;
END BRI_MOVIE_TRG;
.
RUN;


INSERT INTO Movie (name,released) VALUES ('The Lion King',1994);

Oracle’s syntax uses the sequence name with .NEXTVAL or .CURVAL.

Future Directions

I would like to see a SEQUENCE implementation with MySQL (whether official or unofficial). I’m sure some enterprising person in the community already has one. Database abstraction layer systems would also most likely have implementations. I liked the PostgreSQL Syntax for ease of use with the following commands.

  • NEXTVAL(‘sequence’);
  • CURRVAL(‘sequence’);
  • SETVAL(‘sequence’,value);

Wanting something and doing something about it are two different things, so here is what I wiped together to demonstrate a possible implementation. It needs a lot more work in appropiate error handling. transaction management, testing and performance analysis, however it shows the options of one possible implementation.

currval

DROP TABLE IF EXISTS sequence;
CREATE TABLE sequence (
name              VARCHAR(50) NOT NULL,
current_value INT NOT NULL,
increment       INT NOT NULL DEFAULT 1,
PRIMARY KEY (name)
) ENGINE=InnoDB;
INSERT INTO sequence VALUES ('MovieSeq',3,5);
DROP FUNCTION IF EXISTS currval;
DELIMITER $
CREATE FUNCTION currval (seq_name VARCHAR(50))
RETURNS INTEGER
CONTAINS SQL
BEGIN
  DECLARE value INTEGER;
  SET value = 0;
  SELECT current_value INTO value
  FROM sequence
  WHERE name = seq_name;
  RETURN value;
END$
DELIMITER ;

Some Testing:

mysql> SELECT currval('MovieSeq');
+---------------------+
| currval('MovieSeq') |
+---------------------+
|                   3 |
+---------------------+
1 row in set (0.00 sec)
mysql> SELECT currval('x');
+--------------+
| currval('x') |
+--------------+
|            0 |
+--------------+
1 row in set, 1 warning (0.00 sec)
mysql> show warnings;
+---------+------+------------------+
| Level   | Code | Message          |
+---------+------+------------------+
| Warning | 1329 | No data to FETCH |
+---------+------+------------------+
1 row in set (0.00 sec)

What was interesting was I originally used a cursor, as below, but the results for passing an invalid argument (basic boundary testing), returned a SQL error while the above implementation returned a more manageable warning.

 DECLARE c CURSOR FOR
    SELECT current_value FROM sequence
    WHERE name = seq_name;
  OPEN c;
  FETCH c INTO value;


mysql> select currval('x');
ERROR 1329 (02000): No data to FETCH

Indeed the Apache Object Relational Bridge Sequence Manager section shows a very cool syntax for MSSQL.

UPDATE TABLE SET @MAX_KEY = MAX_KEY = MAX_KEY + 1

UPDATE table SET var = column = value which effectively allows you to eliminated the need for a seperate UPDATE and SELECT for this type of operation.

nextval

DROP FUNCTION IF EXISTS nextval;
DELIMITER $
CREATE FUNCTION nextval (seq_name VARCHAR(50))
RETURNS INTEGER
CONTAINS SQL
BEGIN
   UPDATE sequence
   SET          current_value = current_value + increment
   WHERE name = seq_name;
   RETURN currval(seq_name);
END$
DELIMITER ;

mysql> select nextval('MovieSeq');
+---------------------+
| nextval('MovieSeq') |
+---------------------+
|                  15 |
+---------------------+
1 row in set (0.09 sec)

mysql> select nextval('MovieSeq');
+---------------------+
| nextval('MovieSeq') |
+---------------------+
|                  20 |
+---------------------+
1 row in set (0.01 sec)

mysql> select nextval('MovieSeq');
+---------------------+
| nextval('MovieSeq') |
+---------------------+
|                  25 |
+---------------------+
1 row in set (0.00 sec)

setval

DROP FUNCTION IF EXISTS setval;
DELIMITER $
CREATE FUNCTION setval (seq_name VARCHAR(50), value INTEGER)
RETURNS INTEGER
CONTAINS SQL
BEGIN
   UPDATE sequence
   SET          current_value = value
   WHERE name = seq_name;
   RETURN currval(seq_name);
END$
DELIMITER ;

mysql> select setval('MovieSeq',150);
+------------------------+
| setval('MovieSeq',150) |
+------------------------+
|                    150 |
+------------------------+
1 row in set (0.06 sec)

mysql> select curval('MovieSeq');
+---------------------+
| currval('MovieSeq') |
+---------------------+
|                 150 |
+---------------------+
1 row in set (0.00 sec)

mysql> select nextval('MovieSeq');
+---------------------+
| nextval('MovieSeq') |
+---------------------+
|                 155 |
+---------------------+
1 row in set (0.00 sec)

Other References

Ingres CREATE SEQUENCE Page 105.
PostgreSQL CREATE SEQUENCE
Apache Object/Relational Bridge – Sequence Manager – Subproject of The Apache DB Project
MySQL CREATE PROCEDURE

Win4LinPro Trial

In my continuing effort of finding a suitable complete Windows environment under Linux for those small occasions when you just can’t ditch it, I’ve tried Win4Lin. You can download 2 week evaluation copy from the website.


$ su -
$ wget http://www.win4lin.com/administrator/downloads/Win4LinPro-6.2.5-01.i386.rpm
#wget http://www.win4lin.com/index.php?option=com_remository&Itemid=76&func=download&filecatid=3
$ rpm -ivh Win4LinPro-6.2.5-01.i386.rpm


Preparing... ########################################### [100%]
1:Win4LinPro ########################################### [100%]
14 day(s) left in evaluation period.
Creating images... done.
ERROR: Module kqemu does not exist in /proc/modules
Host architecture: i686
KQEMU package: /opt/win4linpro/etc/kqemu.tar.gz
NOTE: KQEMU is Copyright 2005 by Fabrice Bellard.
Win4Lin, Inc. is an authorized distributor of this code.
Building KQEMU module in /tmp/.build_kqemu-24886
Please see the file /tmp/build-kqemu.log in case of error
Installing KQEMU module in /lib/modules/2.6.9-22.0.1.ELsmp/misc
Loading KQEMU module
Starting Win4Lin Pro: [ OK ]

Need to load a Windows XP CD. Note: It must have SP1 or SP2. An older XP CD did not work.


$ loadwinproCD
13 day(s) left in evaluation period.
You may use Win4Lin Pro without entering a license code
until the evaluation period ends. If you already have
a full license code, you may enter it now.
If you do not have a license code, you may enter it at
any time in the future by running (as root):
/opt/win4linpro/bin/ask_license.sh
Would you like to enter your license code now? (y/n) n
Checking CDROM...
Media found: Windows XP Professional (Service Pack 1)
Loading installation files... please wait...
.................................................
Loading CDROM... please wait...
..................................................
Windows CDROM load complete.
$ exit
# must not be be root to continue


$ installwinpro
Selected winxppro to install by default
installwinpro: installation details:
Target directory: /home/rbradfor/winpro
Windows version: winxppro
Guest image size: 4G
mergepro-gowimg: created /home/rbradfor/winpro/GUEST.IMG, size=4G
Enabled KQEMU acceleration
(mergepro-core)
Partitioning guest image...
Guest image partitioning complete.
Enabled KQEMU acceleration
(mergepro-core)
Formatting guest image...
QuickFormatting (only flushing metadata)
Format complete.
4,293,563,392 bytes total disk space
4,293,563,392 bytes available on disk
4,096 bytes in each allocation unit.
1,048,233 allocation units available on disk.
Volume Serial Number is 0D28-140E
Guest image format complete.
installwinpro: launching guest installation...
Enabled KQEMU acceleration
mergepro-gfx: LANG not found, defaulting to: en-us

This then provided the normal windows installation in a window, the only requirement was a serial number, so they have definitely streamlined the Windows installation. Then to run windows.


$ winpro

I did have an installation problem, where I got a CD error of not being able to access a file. Now given that Win4Lin precopied the entire CD, I’m not sure what to make of it, I had to skip the file, and it did not appear to affect the installation. I’ve got a screenshot to upload sometime.

So far, all I’ve really done is run it, I found that it was very slow. What was great, that on the very clean desktop the only icon was a mapping the Linux user home directory. I tried to install Photoshop CS, about the only thing I really want under Windows until I become better with Gimp. Problem was it promptly told me I needed 7GB of space, which I didn’t have. Not sure if I can easily add additional diskspace. Guess I’ll have to RTFM.

XP January Meeting

The Brisbane XP Group met yesterday for a presentation by Dr Paul King of Asert on the book Sustainable Software Development : An Agile Perspective.

I found it a good time to get a collective opinion and review of the techniques and methods we are moving towards in Software Development. Indeed one key point better describing Pair Programming has been added to my upcoming conference presentation Overcoming the Challenges of Establishing Service and Support Channels. I’m hoping Paul makes his notes available as a review of this book, that I will also mention in my presentation.

In Review, this is some of the key points I got from this presentation.

  • Software gradually degrades over time, and will become a maintenance nightmare
  • Successful software will be changed again and again
  • The IT industry has a problem historically with credibility

So the goal is to move towards Sustainability. Some of the points mentioned by Paul were:

  • Continual refinement
  • A working product at all times (not just working software)
  • Value Defect Prevention over Defect Detection
  • Additional investment and emphasis on design

On point I struggle with is Pair Programming. I don’t struggle with the concept, it’s great and really works. The struggle is selling Pair Programming as a core XP Principle. Some good points of discussion lead to a better angle.

  • Pair Programming – should be de-emphasised as a key point
  • By selling Defect Prevention and using Continuous Code Reviews as one method of implementing this
  • Continuous Code Review are achieved with Pair Programming

Much easier. The key point is management understands the term Code Reviews, and if you can show the effect of Defect Prevention on support costs, using Pair Programming, Refactoring and other techniques, your sales pitch will be easier.

Also for reference, the book Software Craftsmanship: The New Imperative was mentioned as a book with similar ideals. A third recommended reading book that was mentioned at the meeting was The Pragmatic Programmer: From Journeyman to Master.





Displaying Images as Text

Well, sometimes you just have to have some fun. Not sure of a pratical use yet. Check out these links.

http://c6.org/toogle/index.php?phrase=Linux+penguin
http://c6.org/toogle/index.php?phrase=marvin
http://c6.org/toogle/index.php?phrase=sunset
http://c6.org/toogle/index.php?phrase=daffy+duck
http://c6.org/toogle/index.php?phrase=polar+bear
http://c6.org/toogle/index.php?phrase=flower
http://c6.org/toogle/index.php?phrase=tulip
http://c6.org/toogle/index.php?phrase=arabx.com.au

PAE Support

Well, learnt a new one today. After a recommendation from a collegue about trying Parellels as a VM tool to run on Linux enabling some more virtual OS’s. (a.k.a. Like VMWare), I eagerly downloaded the rpm and installed. Being caught a bit lately I took the liberty of reviewing the Parallels Workstation 2.0 Installation Guide For Linux as part of the process. Pre-reqs this, ok, run this ok, supports RHEL4 that’s good. Then run the following, accept licence and continue. Fine.


Parallels-config

So I flip to the next page of the manual (pg 13), and there in big letters is:


Troubleshooting
If you receive “Can not configure!” or “Can not compile!” message (note that you should have kernel sources installed), you can view log file and try to fix problem yourself. The log file is named /usr/lib/Parallels/comp.log.< log number>.error
If log contains the following sentence: “Your kernel is compiled with PAE (Physical Address Extension)”, it means that currently running kernel supports PAE. Unfortunately the current version of the Parallels Workstation does not support PAE. To fix the problem:

  • if you have another kernel configuration compiled without PAE support, boot in this kernel.
  • if not, recompile your current kernel to create another one that does not support PAE. Please note that recompiling kernel may lead to unstable work of your primary system. For instructions on how to do this please refer to your primary OS system administration guide or consult your system administrator.
  • run the Parallels Workstation configuration script again.

Well it only took 30 secs for the install to fail. Here I am thinking, given the nightmares I’ve had lately compiling and installing software do I even bother looking.


/usr/lib/Parallels/Drivers/drvMain/vmmodule.h:44:3: #error "Your kernel is compiled with PAE (Physical Address Extension)"
/usr/lib/Parallels/Drivers/drvMain/vmmodule.h:45:3: #error "The current version of the Parallels Workstation does not support PAE mode."
/usr/lib/Parallels/Drivers/drvMain/vmmodule.h:46:3: #error "(Processor type and features -> High Memory Support) - must be not more than 4GB"

Well, made a post on CentOS 4 forums, but I had mixed response with this site.

Anybody know how to recompile a kernel with PAE support? Doesn’t seem a great topic on google.

BTW, in case you missed it PAE is Physical Address Extension.

To enum or not to enum?

I’ve never used database columns that embedded defined valid values within the schema definition. Within MySQL there are 2 definitions, ENUM and SET. There are a few reasons why, but first an explanation of these data types.

In summary, using the MySQL Sample Database.

CREATE TABLE film (
film_id INT UNSIGNED NOT NULL AUTO_INCREMENT,
...
rating ENUM('G','PG','PG-13','R','NC-17') DEFAULT 'G',
special_features SET('Trailers','Commentaries','Deleted Scenes','Behind the Scenes') DEFAULT NULL,
PRIMARY KEY (film_id)
)ENGINE=InnoDB DEFAULT CHARSET=utf8;

So from this, the following commands allow you to inspect this information via mysql.


DESCRIBE film;
SHOW COLUMNS FROM film LIKE 'rating';

With the introduction of the INFORMATION_SCHEMA in MySQL 5, a more traditional method using a valid SELECT statement.


SELECT COLUMN_TYPE
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_SCHEMA='sakila'
AND TABLE_NAME='film'
AND COLUMN_NAME = 'rating'

This is great, however you have to know a lot of information, the table_name and column_name ok, they are fixed, but now the table_schema information is required to be known by the application.

Why I don’t use these?

Historically, there have been 3 reasons why I’ve never used ENUM.

  • First, it’s not standard SQL (at least to my knowledge), and historically hasn’t been consistent between Database products
  • More importantly as data, it’s terrible to maintain easily, as it’s within the table structure
  • The management of valid values within an application, and the need to manage this data dynamically.

Now, I’m not certain if within MySQL there are any funky ways to manage this type of information. I’d welcome comments on what people do. However from my previous experience, this is my method of implementation.

All information of this type, I refer to as a code. I implement this within a common table, defined as reference_data (based historically on the Oracle cg_ref_codes). Here is a cut down version of my definition for simplicity purposes.


CREATE TABLE reference_data(
id INT UNSIGNED NOT NULL AUTO_INCREMENT,
type VARCHAR(20) NOT NULL,
code VARCHAR(20) NOT NULL,
value VARCHAR(100) NOT NULL,
PRIMARY KEY (id)
)ENGINE=InnoDB DEFAULT CHARSET=utf8;

Of course, I’ve extended this basic structure to include ordering of data per type, default value of a given type, grouping of types (for larger systems) and a status to manage historically codes no longer valid for new data.

Within tables, I define all codes with an extention of _code so as to clearly identify the content of a given column.

So, for this example I would add data such as:


INSERT INTO reference_data(type,code,value) VALUES
('rating','G','General Audience - All Ages Admitted'),
('rating','PG','Parental Guidance Suggested. Some Material May Not Be Suitable For Children'),
('rating','PG-13','Parents Strongly Cautioned. Some Material May Be Inappropriate For Children Under 13'),
('rating','R','Restricted. Under 17 Requires Accompanying Parent or Adult Guardian'),
('rating','NC-17','No One 17 And Under Admitted');

This for me provides the following benefits:

  • The Data is data, meaning it can be easily modified via normal application and sql functionality
  • Additional information such as a more general description can be provided and presented to end users
  • Additional attributes non shown in this example, provides ordering support, and a default value
  • The same set of data for a given type can be used in multiple tables

Now, the clear downside of this level of functionality is that the data integrity that is managed by ENUM at the database level has now been lost. You need to temper this with the level of access to your database, and you can easily implement via Database Triggers this level of security. However, I think that clear ability to tie the required values for an ENUM column to easy accessibility via an application, and allow for easy management is clearly a strength.

I’m yet to be convinced otherwise.

Playing .avi files under Linux

Got myself some .avi files but didn’t have a player under CenOS 4.2. MPlayer would appear to rate well as a media player for linux.


yum install mplayer
yum install lame


# Unfortunately mplayer while registered could not be downloaded, tried the hard way
wget http://rpm.greysector.net/yum/4/i386/mplayer-1.0pre7try2-2.i386.rpm
rpm -ivh mplayer-1.0pre7try2-2.i386.rpm
# lame wasn't needed via the manual process, but just in case
#wget http://rpm.greysector.net/yum/4/i386/libmp3lame-3.96.1-4.i386.rpm

Surprisingly, it’s not a GUI interface, but mplayer on the command line is more then adequate.

Support for Technology Stacks

As part of my next conference presentation Overcoming the Challenges of Establishing Service and Support Channels I’ve been struggling to find with my professional sources, any quality organisations that provide full support for a technology stack, for example a LAMP stack, or a Java Servlet stack.

Restricted to searching via online, I’ve been impressed by what I’ve found at Spike Source www.spikesource.com. An organisation with an experienced CEO, well known in the Java Industry. They certainly have all the buzz words covered in their product information.

Benefits of their SpikeSource Core Stack.

  • Fully tested and certified
  • Installs in minutes with integrated installer
  • Enterprise-class maintenance and support available
  • Vendor neutral
  • Horizontally and vertically scalable

SpikeSource offers three prebuilt configurations that can have you up and running in around ten minutes. These configurations comprise the following component choices:

  • LAMP Stack – for Websites with dynamic database-driven content written using Perl and PHP.
  • Servlet Stack – for dynamic Websites written using Java-based Web technologies such as servlets.
  • J2EE Stack – for Web applications that separate Web interface and application logic using Java Servlets and Enterprise JavaBeans.

Supported Platforms. What’s of interest here is RHEL, SuSE as well as Fedora Core 3. In line with for example Oracle software running under Linux.

What’s interesting, is they have MySQL 4.1.14 in their spikesource stack (1.6.2), so they are quite some months behind here. Especially now that MySQL 5 has been available 3 months now. Not only just stack technology, their infrastructure supports a large number of open source products and appears to provide infrastructure via a community to enhance the product offerings within this stack. The Spike Developer Zone Components List provides a long list of products.

Their release notes provide good instructions, in particular what configuration was used in the building of the software. For example, here is the MySQL Release Notes, MySQL Quick Start Guide, MySQL Troubleshooting Guide

They talk about testing, where Core Stack Testing provides more details here.

They also claim to provide VMWare Community Virtual Machine that can be run via the free VM Player on any system without having an effect on an existing system. This is indeed impressive, however it doesn’t seem available. There are many other installations available at the VMWare site.

I’m interested to see what else existing in the marketplace for a fully supported technology stack, rather then support of individual components (e.g. RedHat for Linux, MySQL AB for MySQL, JBoss for a servlet container)

In reading comparisions, there is reference also to Source Labs – www.sourcelabs.com. Anybody that can offer recommendations that I can research would be great.

Book Review – Beyond Java

Well the title got me when I decided to purchase this book “Beyond Java – A glimpse at the Future of Programming Languages”, however perhaps it should have been titled “Why to move from Java to Ruby” as the book for a good portion is an explanation of how Ruby solves the problems that Java has and the direction Java is moving. While the book did describe where Java was, and the future limits and what to look at Beyond Java, the high use of Ruby to describe these overwhelmed the book. In fact, only the last chapter of 20 pages gave an comparison of “Contenders” as the chapter title described other then scant descriptions

Initially I lost count of the number of times information regarding C++ was repeated in the book, and how Java got it’s great penetration from the C++ community. I almost put the book down after the first few chapters, it was highly repetitive.

However, given my increasing interest in Ruby I was able to work though this. I could see a Java developer that has already discard Ruby as a fad to put this book down. In fact, as a Ruby reference it provided some good tips, again strengthening my comment of including Ruby in the title.

Overall an interesting read, however for a small book it could have offered a lot more.

On the same topic, some interesting points in the article The Problems With Java.

Downgrading a MySQL schema from 5 to 4

Why oh why would you want to do this. Well it my case, I’ve committed to developing a web application using MySQL 5 features, knowing that I had to upgrade my production server from 4.0

Well as part of doing this, I hit a stumbling block. My current production web server runs RedHat 7.3, and even with all the latest rpm updates, it does not have glib 2.3 which is required for MySQL 5. I’m no guru, but trying to upgrade from Redhat 7.3 to 9, to at least get these rpm is not an easy process. I’m not confident to try to compile glib 2.3 and all it’s dependancies on a production server, nor is it possible to recompile MySQL down (I suspect not). All just too many variables. It appears the time has come to scrap it and work with a more current RedHat Enterprise Linux version. Down side, the 25 web sites are not going to be too happy.

Anyway, as an interim to at least move forward as much as possible I dowgraded the provided schema.sql to at least run under 4.0. We all know about new features in 5.0, so this is just really the same list, but it was worthwhile documenting.

  • Remove views
  • Remove triggers
  • Remove DELIMITER syntax
  • Change ENGINE= to TYPE=
  • Change DROP SCHEMA to DROP DATABASE
  • Change CREATE SCHEMA to CREATE DATABASE
  • Remove DEFAULT CHARSET=utf8 from table definition
  • Remove DEFAULT TIMESTAMP from TIMESTAMP columns (NOTE: Need to verify the column is the first TIMESTAMP column)
  • Change BOOLEAN to TINYINT(1)

Update:
Friend and collegue Morgan points out that if I was dumping a schema with 5.1 you could use the syntax mysqldump --compatible=name. Unfortunately in my case the schema was coming from a third party, however now I know this syntax, I can always ask for it.

Update 2:
I know in earlier versions that you could get a MySQL product installation, and also get one compiled with glib23. I just figured that over time the older glib22 were dropped, and were not available with 5.1. It seems domestic blindness has now hit me in computer software. I did review the MySQL 5.1 downloads page, and did not see lower in the page mixed in the 30 to 40 downloads and option to download a glibc22 version. Thanks Arjen for pointing that out.


Domestic Blindness
“The inability to find a common object that is right in front of your face.”