Hazel Nuts

A programming gem

LL — Tue, 14 Jun 2005 22:32:49 +0000

While being a little offset because of some private matters, I began to learn (in what’s left of my free time) a new programming language.

The language is Ruby, developed by a single japanese guy. It seems to be something between Smalltalk and Perl - with quite simple syntax, full object orientation, exception handling and what not.

First impressions - the syntax is still a bit disturbing for me, some concepts are unfamiliar, some things seem to be overly complicated - but generally nicer than Perl, and closer to shell scripts than Python. Promising.

Some additional resources can be found on Ruby Central. Another thing worth taking a look is Rails, a open-source web framework written entirely in Ruby.

For the moment however, I’ll stick with Python as a scripting language of choice - but you never know…

One of the reasons I don’t like Oracle

LL — Sat, 28 May 2005 23:00:52 +0000

Have you ever worked with clobs (or blobs, doesn't matter) in your database? Well, I did. And while it generally poses no problems in any database I know, in Oracle it is cumbersome at best. Would you expect this to work?

INSERT INTO TABLE (id, clob_col)
VALUES (1, 'A very long line...');

Hmm, I would. For some misterious reasons it works everywhere - but not in Oracle! If the line is longer than 2499 characters (in version 8.something, maybe it's different in other versions), all you get is a friendly error message. What to do? Write a program, say, in Java. There is about 2.7 million ways to do it, of which this is one of the easiest:

//say we already have a connection and a statement object
//this is needed. I don't know why.
connection.setAutoCommit(false);
//insert row with empty clob
statement.executeQuery("insert into table(id, clob_col) "+
"values (667, EMPTY_CLOB()) ";
//get the Clob object back from DB
ResultSet rs = stat.executeQuery("select clob_col "+
"from table "+
"where id = 667 for update");
rs.next();
Clob c = rs.getClob(1);
//and finally - this is it!
((oracle.sql.CLOB)c).putString(1, "OMG");
//commit the whole thing
conn.commit();

So much fun...

Leave your mark

LL — Tue, 24 May 2005 15:42:08 +0000

(click to see full picture)

The painting has been done on the wall near the entrance of the main subway station in Warsaw. This is the third attempt- the first one was silly, the second outrageous, and this one... I don't know what to think of it. Looks nice, but I guess that every owner of anything bigger than a waterproof marker will just crave to leave his mark. Let me be the first one:

BibCenter available for download

LL — Sat, 21 May 2005 23:38:46 +0000

First version of BibCenter- a bibliographical reference manager- is available for download. It currently features:

database- driven information storage
a BibTeX compliant information set
export to BibTeX file
customizable additional fields

Get the software it in the downloads section.

Processing rev. 90 available

LL — Thu, 19 May 2005 23:52:09 +0000

A new revision of Processing (beta version) is available with some new features like full screen graphics, a more complete renderer and better opengl support.

Get the latest revision at http://processing.org/download/.

Java BBCode parser

LL — Thu, 19 May 2005 23:44:53 +0000

I assembled a simple BBCode parser in Java. It might be useful, e.g. in a custom bulletin board. It is still a bit incomplete, but will probably improve over time.

You can download it in the downloads section.

Three pointers on personal data in databases

LL — Tue, 17 May 2005 11:45:57 +0000

Pointer 1

A mistake commonly made by database designers, is to represent name or surname with single columns. This approach, however easy and natural, clearly breaks the first normal form, because the relations in the database become non- singular when multiple names occur. This makes searching more difficult, because many combinations and '%'s have to be used.

Consider for example, a person with 3 names - "William Henry Bob", and a search string "William Bob". In the "popular" approach, the name of the person is put into single column, say we call it "name". The SQL query will look something like this:

SELECT * FROM people
WHERE name LIKE '%William%'
OR name LIKE '%Bob%';

Simple, but very consuming for the database. You need to run a full table scan, which is not very good when the database is big.

Instead, apply the normalization procedure required by the first normal form:

Now, the SQL query will look somewhat like:

SELECT * FROM people WHERE id IN (
SELECT people_id FROM people_names WHERE names_id IN (
SELECT id FROM names WHERE name IN ('Bob','William')
)
);

This query seems to be a bit more complicated, but - not surprisingly - runs a lot faster.

Pointer two

Because of various errors, names or surnames in the database can be misspelled. A simple search with exact matching is obviously not able to find the misspelled words.

A partial solution of this problem is phonetic indexing. A phonetic index is a coded word hash based on the way a word sounds rather than the way it is spelled (eg. SMYTH or SMITH sound the same, while being spelled differently). There is a couple of phonetic algorithms:

The original soundex algorithm has been invented by O'Dell and Russell in 1918. A variation called "American soundex" has been used by the U.S. census archives. This algorithm is very simple, thus not very accurate;
The metaphone algorithm, developed by Phillips, is more accurate and more complicated;
The Daitch- Mokotoff soundex has been developed by the Jewish genealogists in 1985. It had to be done because the two above algorithms are both based on English language and don't work very well for Slavic or Germanic surnames (eg. Levine vs. Lewin).

Both the soundex and metaphone algorithms are part of PHP standard library. There also exist freely available implementations in Java at Jakarta. Of course many other realisations can be found on the internet.

Pointer three

The longest surname I know in Polish language is "Brzęczyszczykiewicz", which is 20 characters long. In a database, I assume 24 chars for surnames and 18 for names.

No activity

LL — Mon, 16 May 2005 20:21:32 +0000

I'm killing my brain cells with Addicting Games.

Spreading the word - badly

LL — Fri, 13 May 2005 19:31:56 +0000

This almost looks like a sabotage: http://funnyfox.org/

OMG. Is Microsoft behind this?

Won’t play with Scrabble

LL — Thu, 12 May 2005 22:52:13 +0000

I just bought the "Official Polish Scrabble Dictionary". Partly because of hope that this dictionary would be in an open format, so I could use it to write some funny stuff. No, it isn't. Forlorn hope.