Application Configuration

The configuration of an application is represented by a directory. It may contain the following:

rose-app.conf - a compulsory configuration file in the modified INI format. It contains the following information:
- the command(s) to run.
- the metadata type for the application.
- the list of environment variables.
- other configurations that can be represented in un-ordered key=value pairs, e.g. Fortran namelists.
file/ directory: other input files, e.g.:
- FCM configuration files (requires ordering of key=value pairs).
- STASH files.
- other configuration files that require more than section=key=value.
Files in this directory are copied to the working directory in run time.

Note

If there is a clash between a [file:*] section and a file under file/, the setting in the [file:*] section takes precedence. E.g. Suppose we have a file file/hello.txt. In the absence of [file:hello.txt], it will copy file/hello.txt to $PWD/hello.txt in run time. However, if we have a [file:hello.txt] section and a source=SOURCE setting, then it will install the file from SOURCE instead. If we have [!file:hello.txt], then the file will not be installed at all.
bin/ directory for e.g. scripts and executables used by the application at run time. If a bin/ exists in the application configuration, it will prepended to the PATH environment variable at run time.
meta/ directory for the metadata of the application.
opt/ directory (see Optional Configuration).

E.g. The application configuration directory may look like:

./bin/
./rose-app.conf
./file/file1
./file/file2
./meta/rose-meta.conf
./opt/rose-app-extra1.conf
./opt/rose-app-extra2.conf
...

Rose Configuration rose-app.conf

Config file-install-root: Root level setting. Specify the root directory to install file targets that are specified with a relative path.

Config meta: Root level setting. Specify the configuration metadata for the application. This is ignored by the application runner, but may be used by other Rose utilities, such as rose config-edit. It can be used to specify the application type.

Config mode

Root level setting. Specify the name of a built-in application, instead of running a command specified in the [command] section.

See also Rose Built-In Applications.

Config opts: Hardcode an optional configuration to be used by the application. It is generally better to specify optional configurations using the ROSE_APP_OPT_CONF_KEYS environment variable or --opt-conf-key argument both of which work with rose app-run and rose task-run.

Config [command]

Specify the command(s) to run.

Config default= COMMAND

Compulsory:: True

Specify the default command to run.

Config ALTERNATE= COMMAND

Specify an alternate command referred to by the name ALTERNATE which can be selected at runtime.

See the Command Keys tutorial.

Config [env]

Specify environment variables to be provided to the [command] at runtime.

The usual $NAME or ${NAME} syntax can be used in values to reference environment variables that are already defined when the application runner is invoked. However, it is unsafe to reference other environment variables defined in this section.

If the value of an environment variable setting begins with a tilde ~, all of the characters preceding the first slash / are considered a tilde-prefix. Where possible, a tilde-prefix is replaced with the home directory associated with the specified login name at run time.

Config KEY= VALUE: Define an environment variable KEY with the value VALUE.

Config UNDEF

A special variable that is always undefined at run time.

Reference to it will cause a failure at run time. It can be used to indicate that a value must be overridden at run time.

Config [etc]: Specify misc. settings.

Tip

Currently, only UM defs for science sections are thought to require this section.

Config [file:TARGET]

Specify a file/directory to be installed. TARGET should be a path relative to the run time $PWD or STDIN.

E.g. file:app/APP=source=LOCATION.

For a list of configuration options see *[file:TARGET] for details.

Config [namelist:NAME]

Specify a Fortran namelist with the group name called NAME, which can be referred to by a *[file:TARGET]source setting of a file.

Config KEY= VALUE: Define a new namelist setting KEY set to VALUE exactly like a Fortran namelist, but without the trailing comma.

Namelists can be grouped in two ways:

[namelist:NAME(SORT-INDEX)]
- This allows different namelist files to have namelists with the same group name. These will all inherit the same group configuration metadata (from [namelist:NAME]).
- This allows the source setting of a file to refer to all [namelist:NAME(SORT-INDEX)] as namelist:NAME(:), and the namelist groups will be sorted alphanumerically by the SORT-INDEX.
[namelist:NAME{CATEGORY}]
- This allows the same namelist to have different usage and configuration metadata according to its category. Namelists will inherit configuration metadata from their basic group [namelist:NAME] as well as from their specific category [namelist:NAME{CATEGORY}].

These groupings can be combined: [namelist:NAME{CATEGORY}(SORT-INDEX)]

Config [poll]

Specify prerequisites to poll for before running the actual application. Three types of tests can be performed:

Config all-files: A list of space delimited list of file paths. This test passes only if all file paths in the list exist. Accepts globs.

Config any-files: A list of space delimited list of file paths. This test passes if any file path in the list exists. Accepts globs.

Config test

A shell command. This test passes if the command returns a 0 (zero) return code.

This test is run in addition to the all-files and any-files tests for the existence of file paths.

Config file-test

A shell command run for each file path, where {} is substituted for the file path at runtime. This test passes if the command returns a 0 (zero) return code for all paths in all-files and any path in any-files.

If test is not enough, e.g. you want to test for the existence of a string in each file, you can specify a file-test to do a grep. E.g.:

all-files=file1 file2
file-test=test -e {} && grep -q 'hello' {}

At runtime, any {} in the command is replaced with the name of the file (or glob pattern). The example above tests that both file1 and file2 exist and that they both contain the string hello. It is run for each path/pattern, so the example above would be equivalent to:

test -e file1 && grep -q 'hello' file1
test -e file2 && grep -q 'hello' file2

Config delays

The above tests will only be performed once when the application runner starts. If a list of delays are added, the tests will be performed a number of times with delays between them. If the prerequisites are still not met after the number of delays, the application runner will fail with a time out. The list is a comma-separated list. The syntax looks like [n*][DURATION], where DURATION is an ISO8601 duration such as PT5S (5 seconds) or PT10M (10 minutes), and n is an optional number of times to repeat it. E.g.:

# Default
delays=PT0S

# Poll 1 minute after the runner begins, repeat every minute 10 times
delays=10*PT1M

# Poll when runner begins,
# repeat every 10 seconds 6 times,
# repeat every minute 60 times,
# repeat once after 1 hour
delays=PT0S,6*PT10S,60*PT1M,PT1H