Friday, January 08, 2010

MySQL FULLTEXT search and stopwords

If you are trying too look for a word like 'yours' in the TEXT MySQL field you will never get any result. This is because 'yours' is one of the 'stopwords' ignored in the MySQL searches.
Here is the list of all such stopwords


The bad thing is that this behavior does not change even if you search in BOOLEAN mode. This note describes this as a requested fix for some later MySQL release.

Anyway, if you want to search for just ANY word, without taking in consideration the 'stopwords' list of MySQL, add this line

ft_stopword_file=""


in [mysqld] section of your my.cnf file (for XAMPP on Win xamp-folder/mysql/bin/my - no extension)

Next:
a) restart mysql daemon / service
b) rebuild tables indexes, e.g. by the following SQL comand:

REPAIR TABLE `Texts` QUICK

Now you can search for words like "can", "me" and "you" and - what a surprise! - find the records containing them.

Thursday, December 17, 2009

Free antiviruses compared - AVG vs. Avast vs. Antivir

Well, I tried all three of them in the real work environment.
For the last couple of years I was using AVG and it worked good for me.
Once configured, it smoothly run on background doing it's job.

However, for the last 6 months or so I noticed that my Thunderbird (2.0.0.23 at the time of writing) was getting slow. It was slow on emails download, on emails browsing and on emails delivery. I also found that 1 of my 4 Quad CPU cores was at 100%, executing one of AVG child processes. I tried to re-configure AVG or re-install the newest version from the scratch and I looked for anything wrong with my Thunderbird configuration, but was unable to find any clue.

Additionally, I noticed that the latest free AVG 9.0 won't auto-update anymore. It requires manual update, which is obviously a serious problem for busy people. I can not remember that my anitvirus is not updated and AVG 9.0 even does not notify me about that.

So I ended up removing AVG after being their user for a long time. Now the options were Avast! or Antivir.

I installed both in parallel on a test computer, just to let them breath for a while. I am glad I did that since after a few days I removed Antivir. It could be a great piece of software, but it's German creators decided to show you a periodic reminder (daily or may be once in a few days) that the paid version gives you more protection. So the first thing you see when you open your computer is a big pop-up window "buy Antivir". Thank you, I removed this nice auto-reminder software when I got the second "buy now" message.

Thus I am with free Avast! antivirus now. Avast! works good with my Thunderbird, all my email issues gone. Emails download is fast and Thunderbird never freezes up anymore on emails browsing or deletion. Like all the others, Avast! has it's own weird stuff, like the default voice notification 'virus database is updated'. Fortunately, sound can be switched off through Settings->Sounds->Disable Avast Sounds.
A few more annoying things that I yet did not found how to avoid yet include:
- small auto-closing pop up notification window on auto-update. I do not want to see it.
- pop-up if an infected email arrived - I would like Avast! to take a default action (delete the attachment / virus in my case) and never show me any window.

Anyway, at the moment Avast! is the best choice for my environment.

Wednesday, November 11, 2009

Installing ImageMagick on Fedora 8 Linux VM

There was a relatively old version of software installed, so I preferred to install from source instead of 'yum install ImageMagick' who told me 'there is no new versions'. I should be verifying why my repositories say that, I know...

Anyway, it was a very easy task. Here are the simple instructions how to install from source: http://www.imagemagick.org/script/install-source.php#unix

Summarizing that, there are just a few simple commands (you should get root permissions or use 'sudo make install' instead of 'make install' below)

tar xvfz ImageMagick.tar.gz
cd ImageMagick-version
./configure
make
make install


'make' took some time, but 'make install' on my machine failed with:
'Can't locate ExtUtils/MakeMaker.pm' error in Makefile.PL

I solved this by installing the missing package:

yum install perl-ExtUtils-MakeMaker


and running 'make install' again. Running

convert logo: ml.gif


silently created an ImageMagick logo image that can bee viewed in 'ml.gif'

Done

Installing MagickWand for ImageMagick with XAMPP / PHP on Windows

XAMPP comes with ImageMagic. Check you see this section running phpinfo():


imagick
imagick module enabled
imagick module version 2.0.0-rc1
imagick classes Imagick, ImagickDraw, ImagickPixel, ImagickPixelIterator
ImageMagick version ImageMagick 6.3.3 04/21/07 Q16 http://www.imagemagick.org
ImageMagick release date 04/21/07
ImageMagick Number of supported formats: 164
ImageMagick Supported formats A, ART, AVI, AVS, B, BIE, BMP, BMP2, BMP3, C, CACHE, CAPTION, CIN, CIP, CLIP, CLIPBOARD, CMYK, CMYKA, CUR, CUT, DCM, DCX, DFONT, DPS, DPX, EMF, EPDF, EPI, EPS, EPS2, EPS3, EPSF, EPSI, EPT, EPT2, EPT3, FAX, FITS, FRACTAL, FTS, G, G3, GIF, GIF87, GRADIENT, GRAY, HISTOGRAM, HTM, HTML, ICB, ICO, ICON, INFO, JBG, JBIG, JNG, JP2, JPC, JPEG, JPG, JPX, K, LABEL, M, M2V, MAP, MAT, MATTE, MIFF, MNG, MONO, MPC, MPEG, MPG, MSL, MSVG, MTV, MVG, NULL, O, OTB, OTF, PAL, PALM, PAM, PATTERN, PBM, PCD, PCDS, PCL, PCT, PCX, PDB, PDF, PFA, PFB, PGM, PGX, PICON, PICT, PIX, PJPEG, PLASMA, PNG, PNG24, PNG32, PNG8, PNM, PPM, PREVIEW, PS, PS2, PS3, PSD, PTIF, PWP, R, RAS, RGB, RGBA, RGBO, RLA, RLE, SCR, SCT, SFW, SGI, SHTML, STEGANO, SUN, SVG, SVGZ, TEXT, TGA, THUMBNAIL, TIFF, TILE, TIM, TTC, TTF, TXT, UIL, UYVY, VDA, VICAR, VID, VIFF, VST, WBMP, WMF, WMFWIN32, WMZ, WPG, X, XBM, XC, XCF, XPM, XV, XWD, Y, YCbCr, YCbCrA, YUV


If you got it, you have ImageMagick installed. If you do not see this section, install ImageMagick first: go to http://www.imagemagick.org

Next, if you are lazy like I am install a compiled version of MagicWand.
Get the binaries here: http://www.dirk.sh/dirk/magickwand/
I used php-5.2.4_magickwand_q16_st.dll and it seems to work fine.
Alternatively, get the source code from http://www.magickwand.org and compile it by yourself.

Put the DLL into PHP extensions directory. If you are not sure where it is, look into your php.ini file for a line like
extension_dir = "C:\xampp\php\ext\"

There are at least two php.ini with XAMPP for Windows distribution. The correct one is located in your-xampp-installation-folder\apache\bin, e.g. "C:\xampp\apache\bin\php.ini"

Define the new extension by adding to php.ini a line like this:
extension=php-5.2.4_magickwand_q16_st.dll

Restart Apache

Verify that phpinfo() now shows the MagickWand table:


magickwand
MagickWand Backend Library ImageMagick
MagickWand Extension Version 1.0.5
ImageMagick support enabled
ImageMagick version ImageMagick 6.3.5 09/21/07 Q16 http://www.imagemagick.org
ImageMagick QuantumRange (MaxRGB) 65535
MagickWand supported image formats A, ART, AVI, AVS, B, BIE, BMP, BMP2, BMP3, C, CAPTION, CIN, CIP, CLIP, CLIPBOARD, CMYK, CMYKA, CR2, CRW, CUR, CUT, DCM, DCR, DCX, DFONT, DNG, DPS, DPX, EMF, EPDF, EPI, EPS, EPS2, EPS3, EPSF, EPSI, EPT, EPT2, EPT3, EXR, FAX, FITS, FRACTAL, FTS, G, G3, GIF, GIF87, GRADIENT, GRAY, HISTOGRAM, HTM, HTML, ICB, ICO, ICON, INFO, IPL, JBG, JBIG, JNG, JP2, JPC, JPEG, JPG, JPX, K, LABEL, M, M2V, MAP, MAT, MATTE, MIFF, MNG, MONO, MPC, MPEG, MPG, MRW, MSL, MSVG, MTV, MVG, NEF, NULL, O, ORF, OTB, OTF, PAL, PALM, PAM, PATTERN, PBM, PCD, PCDS, PCL, PCT, PCX, PDB, PDF, PEF, PFA, PFB, PFM, PGM, PGX, PICON, PICT, PIX, PJPEG, PLASMA, PNG, PNG24, PNG32, PNG8, PNM, PPM, PREVIEW, PS, PS2, PS3, PSD, PTIF, PWP, R, RAF, RAS, RGB, RGBA, RGBO, RLA, RLE, SCR, SCT, SFW, SGI, SHTML, STEGANO, SUN, SVG, SVGZ, TEXT, TGA, THUMBNAIL, TIFF, TIFF64, TILE, TIM, TTC, TTF, TXT, UIL, UYVY, VDA, VICAR, VID, VIFF, VST, WBMP, WMF, WMFWIN32, WMZ, WPG, X, X3F, XBM, XC, XCF, XPM, XV, XWD, Y, YCbCr, YCbCrA, YUV


You are mostly done now.

Test your installation with the PHP code like this


$resource = NewMagickWand();
MagickReadImage( $resource, 'D:\test.jpg' );

header( 'Content-Type: image/jpeg' );
MagickEchoImageBlob( $resource );


And look here for some more PHP / MagickWand / ImageMagick code snippets.

Also note this nice post, however the links to ImageMagick binaries there do not work anymore.

Friday, June 05, 2009

Inside AdWords: New Interface Thursdays: Exploring the Ads tab

Inside AdWords: New Interface Thursdays: Exploring the Ads tab
Well, I tried the new interface and due to multiple processing errors I meanwhile switched back to the old one, thanks G. :-) it is still available.
Working with Adwords-online this time, I found a strange limitation of maximum 50 ads in the same ad group. I believe it exists in both old and the new UI, seems that the limitation should be at the server side.

I just wanted to put in about 25 ads, each one delivered in 8 Google' banner formats - from 468x60 to 160x600. Yes, I wanted to put them all in the same adgroup but I could not do that because of this stupid 50-max limit.

Google forced me to open new ad groups in the same ad campaign (I never wanted that) and put my extra- banners there. I ended up with 3 ad groups instead of just one because of 50-ads max limitation. (3 but not 4 groups because not all the ads where delivered in all sizes).

Now, I have to manage separately these 3 ad groups - their keywords, bids etc. Obviously, this is not what I want to do - it wastes my time without any return.

So, dear Google, would you please remove this weird 50-max ads limit, letting me to do my job faster? I need it much more than WEB2-like new UI you guys have introduced. I do not say it is baaad, it is just still to young to work properly with all the AJAX involved. Is not it enough that Microsoft treat their users as testers, delivering them new OS that should never leave their lab, now Google is going the same way?

Thursday, May 21, 2009

mySQL FULLTEXT search returns no results: ft_min_word_len

MySQL FULLTEXT search query worked well until I tried to find "boy" in my DB.
No results, and I can see some records with this word directly on my phpMyAdmin screen.
The same repeated with the word "red". It can find, "blue"< "green" and "yellow", but not "red".

Well, after some searching I found that FULLTEXT in mySQL has two server variables (well, it have more of course, but these two are relevant for this issue):
ft_min_word_len = 4
ft_max_word_len = 84

Words less than 4 chars will not be indexed until you change ft_min_word_len to something less than 4 chars.

So, on Redhat/Fedora and many other Linuxes you have to edit /etc/my.cnf
The addition might look:

[mysqld]
ft_min_word_len=1
ft_max_word_len=32

Note [mysqld] header - if it is already in that file, just put two lines woth ft_ under it. It's important! If you put it under another header mySQL will not recognize it. Cost me about 2 hours two figure this stupid thing out since typically "[...]" headers are ignored by the parsing software (used as comments).
But obviously, not in my.cnf and mySQL.

On Windows with XAMPP there is a file called "my" (just "my", now extension whatsoever) in /mysql/bin folder.
It's the same my.cnf, so just add the lines into [mysqld] section there.

Verify that the change worked in phpMyAdmin ->localhost -> Varaibles

The last thing to do to get the things working is rebuilding all FULLTEXT indexes, if you got some data in the DB yet.
With phpMyAdmin you can just click "edit" link next to the FULLTEXT index name, and when the edit screen opens, click "Save" button. You will see that mySQL runs actually 'drop the old index and create the new index" query.

Monday, March 30, 2009

Switching between different Java versions, Java 5 instead of Java 6

Since like many of you I am developing multiple projects simultaneously and have a very heterogeneous environment on my Windows machine, I always looked for a simple way to switch between the latest Java version and the previous one, which I still needed for some older projects.
Once installed, the latest JRE/JDK (currently, Java 6) become the JVM of choice for any app in Windows OS and it was quite tricky for me to force an app to be executed under Java 5.

I saw many posts on the Internet claiming they solved the issue by manually
editing Windows registry and/or replacing soem files in Windows/system32 directory - never wanted to do that.

This post is probably the closest to what I wanted to do, however the code from this post did not work 'as is' in my batch files.

Looking at java.exe command options in Java 6 I found an amazing feature:
-version:

This is what the command line java says about it:
-version:
require the specified version to run

Sounds just what I needed! So I run it like this:
>java -version:1.5.0_17 -version

And voila! - here is what says my Java 6 now:
java version "1.5.0_17"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_17-b04)
Java HotSpot(TM) Client VM (build 1.5.0_17-b04, mixed mode)


Of course if I run it with just -version (no ":" and no alternative JRE/JDK), it keeps saying:
> java -version
java version "1.6.0_12"
Java(TM) SE Runtime Environment (build 1.6.0_12-b04)
Java HotSpot(TM) Client VM (build 11.2-b01, mixed mode, sharing)