Installing Niagara on Windows NT/2000
The niagara query engine requires
JDK 1.2 or later to compile. This procedure was tested on Windows
NT, but it should also work on Windows 2000 and might work on Windows 95/98.
Obtaining and Unpacking
Use your favorite program to unzip
niagara_source.zip into your favorite directory, preserving the directory
structure in the archive.
The top level directory should contain src, lib and
demo directories and a batch file called make.bat
Use notepad to edit make.bat There are two
clearly marked variables that may need to be modified:
JAVAHOME <== Set this to your jdk directory.
The default is c:\program files\jdk1.2.2
COMPILEFLAGS <== Set your compile flags here.
The default is -g
Save and execute make.bat
There may be some warnings during the compilation,
and those are ok. If something goes wrong, the compilation will stop
at the first error.
If everything goes well, there should be 4 new
batch files and a jar file in the main directory. They batch files
are buildIndex.bat, runSearchEngine.bat, runQueryEngine.bat and runClient.bat.
Note that these files are recreated every time you run make.bat, so
all changes you make to them will be lost.
Running the server
Ignore this section if you only wish to run the client
and connect to an already exisiting Niagara server.
Before you run the Niagara server for the first
time, you must first build the indexes on the XML data.
execute buildIndex.bat (only needs to be done
whenever you want to build/rebuild indexes!)
The indexes will be built over the urls specified
The the behavior of the Search Engine can be controlled
Running the client
by default the client connects to www-db.cs.wisc.edu.
If you want it to connect to a different Niagara server, modify runQueryEngine.bat
Questions or comments can be sent to
Arthur Kantor or to any of the people