My Postgres Story: Internationalization: obartunov — LiveJournal

I used Postgres since 1995 ( [PG95] mailing record has lower than 400 subscribers !) for my scientific initiatives and working with numbers and english strings was easy, however in 1996 I signed as much as make a searchable database of digital archive of “Uchitelskaya gazeta” ( the favored newspaper for academics group). It was my moonlight work to assist me develop youngsters. The entire challenge was sponsored by West basis, if I recall correctly, our workplace was positioned in Vetoshny Lane, close to the Learn Sq., I had loads of enjoyable working there with sensible folks. I claimed to be an skilled in Postgres and did not see any downside with loading the archive and writing a number of SQL queries 🙂 First, I shortly discovered that cyrillic letters reworked to one thing unreadable. The issue was that Postgres operated with ASCII solely, which is not shocking given his roots in English world, so virtually it was not possible to make use of any nationwide encodings like russians koi8-r, cp-1251, cp-866, and so forth. That point I already participated in internationalization of perl and search engine glimpse, so I knew that principally I would like to switch ‘char’ definiton to ‘unsigned char’, add name setlocale(LC_ALL,””) and use capabilities from locale api. The truth is getting a bit extra difficult than I anticipated, so it took a month to get a working Postgres on Slackware. The second downside I discovered is that postgres works very gradual with textual content, it was not solely due to locale stuff — strcoll is 10 occasions slower that strcmp, however since Postgres used full scan — learn each row of a desk and performs gradual textual content operation. This downside was addressed by me and Teodor in 2000, once we developed OpenFTS, contrib/intarray (see this historical document for particulars), which was finally transforms to built-in full-text search (FTS), which we offered on the PostgreSQL Anniversary Assembly in Toronto, 2006.
For years I didn’t understand the significance of this primary patch, solely lately I understood that it made Postgres an worldwide open supply database. Many builders proceed this work, Tatsuo Ishii launched help of multi-byte encondings, Peter Eisentraut made loads of enhancements to internationalization of Postgres.
I select to make use of Postgres over Oracle in 2000, once we developed the Rambler portal (the most important portal in Russia, the hundreds of thousands of customers) , since I believed that we may do every little thing having sources and good help from mailing record. I deal with the strain of Rambler investor and high administration, who wished Oracle to move an audit by PWC. Only for reference, we began with Postgres 6.5.2 (no WAL!) progressively including GiST framework, contrib/intarray, OpenFTS, tsearch v1. Because of Rambler, all these work grew to become finally a part of Postgres !