Useful tools in the iLand world
Table of contents
iLand uses the cross platform C++ toolkit Qt. Qt is available for Windows, Linux and Mac platforms (along several mobile platforms like Symbian, Maemo, and others) - so you are not locked in with regard to operating systems. The Qt framework is open source and offers lots of functionality beyond the graphical user interface.
iLand uses the SQLite (http://www.sqlite.org/) database. This is a leight-weight open source database that requires no installation, is cross platform and has the reputation to be very fast.
SQLite stores the all tables of a database within one single file. This approach could get quite handy for the storage of model outputs.
There are several open source tools available that help the developer / user with the handling of data. Especially important are the import / export features.
Over time, I tested a number of packages:
- Sqliteman (http://sqliteman.com/). This is my favorite: It is written in Qt, and seems to offer all which is needed.
- SqlLite Studio (http://sqlitestudio.one.pl). This tool is nice, because it requires no installation (just a single Exe-File of ca. 2.5MB), but lacks the ability to import data (and has a slightly old-fashioned look and feel). One advantage is the convenient editing of table rows (dubbed as Form view).
- DB Browser for SQLite: a nice and actively maintained tool available for Mac/Windows. Functionality is on par with Sqliteman.
- a new addition to the list: a Firefox-plugin to edit SQLite databases(!). It supports renaming of columns, export to CSV and much more. Though originally designed to view/edit Firefox related data (yes, Firefox stores all its stuff (bookmars, history, ...) in a SQLite database), it does a good job for iLand data as well. This is a nice example of the open source biotop...
A ODBC driver for the DB is also available: http://www.ch-werner.de/sqliteodbc/, some discussion on http://www.sqlite.org/cvstrac/wiki?p=SqliteOdbc. This should allow access from other tools e.g. R (statistics package).
Just recently, I stubled upon the RSQLite Package for R - this should allow direct access from R...
The SQLite database used through the QSqlDatabase abstraction layer of Qt shows awesome performance. It is very fast when writing large output tables (a test table with just 3 or 4 columns is filled with 100000 rows/sec), but also fast at reading. Here the test logical test case is climate data - due to iLand's use of daily data. Here a query of 100 years (i.e. approx. 36000 rows) takes about 0.5 seconds.
One performance issue, though, is the use of transactions. If DML statements (insert, update) are not explicitly surrounded by a transaction (i.e. begin transaction and end transaction), the execution is suspended until the data is actually written back to disk.