The files belonging to a project are organized according to a specific set of conventions which are described here.
The files belonging to a project are strictly organized; no files
may be present in the project's home directory except system
configuration files (these generally begin with a period
.) and so are not normally visible when listing the
project home directory. Several groups of directories contain static
and computer-generated project content, and all of these begin with
digits. Any directory whose name does not begin with digits is by
definition a sub-project.
Most project workers will access project files from the project
home directory. Thus, when we refer to a
00atf, that would be accessed via Emacs using
PROJECT would be replaced by the actual project
The programs behind Oracc, however, access project files is from
the top-level Oracc installation
directory. This matters to you because when a program refers to,
00atf directory in an error message, it will probably
call it something like
This access route is the reason why project names must all be four characters or more in length: Oracc reserves names of three or less letters for system files and directories.
This is also the reason why project names must be in all lowercase: for simplicity and predictability all Oracc directory names consist only of lowercase letters, in some cases prefixed by a sequence of digits.
All project content which can be edited by hand is contained in one of the directories
00. The current set of these directories is:
00cat; note that projects which use the CDLI main catalogue directly have their metadata created dynamically when the project is built, and do not have any catalogue files in the 00-tree
00lib; text lists live in
00mapdirectory which contains links to the files in
00atfin a more user-friendly format
During the build process, a number of XML versions of the
text-based sources are generated, and these are placed in the
01bld. Each text has its own directory
01bld, named according to its ID. When one project
borrows data from another via proxying, it is this ID-named directory
from which the proxied data is taken.
Unmerged glossary data harvested from ATF files is stored in
There is also a working directory named
Both of these directories actually live in the variable part of the filesystem, by convention in /var/local/oracc. The entries in the project home directory are links.
There are several different kinds of public data besides the raw build results.
01bld these directories are also links to the Oracc directory
00prefixes so that they occur first in the project data in a simple sort
Subprojects work exactly like projects; the only difference is that their home directory is a project or subproject directory, rather than the home directory of the project.23 Jul 2014
Steve Tinney & Eleanor Robson
Steve Tinney & Eleanor Robson, 'Project files', Oracc: The Open Richly Annotated Cuneiform Corpus, Oracc, 2014 [http://oracc.museum.upenn.edu/doc/help/managingprojects/projectfiles/]