a number of large XML files available as test cases

[ari@cogsci.ed.ac.uk (K. Ari Krupnikov)]

20 "data-oriented" files, from 301K to 122M for stress-testing your
applications are available at http://www.ltg.ed.ac.uk/~kari/dafif/

I produced the files as an XML version of Digital Flight Information
File (DAFIF) produced by NIMA - US National Imagery and Mapping Office
(formerly known as Defense Mapping Agency - DMA).

The original data are a relational database dump. In the XML version,
every element represents a row in a table; attributes represent
values, if a value is NULL, the corresponding attribute is absent;
there is no character data. Foreign key relationships are modeled as
child elements.

This version reproduces the original dataset exactly, with no
modifications to structure or content except for stripping NULLs. Many
improvements to its format are possible for any application I can
think of -- NF1 could be a good start. However, these files can be
useful in testing precisely because they represent a "real life"
dataset as it appears in the wild.

The tab-delimited source is available from NIMA at
ftp://164.214.2.112/dafif/dafif_0303_ed7/DAFIFT.ZIP

For those interested, NIMA's documentation for the original format is
here: ftp://164.214.2.112/dafif/dafif_0303_ed7/DOCS/ED7/


Ari.